Pöördumine

Doc#: OPZ-021-05 Page 1 of 2

EC Declaration of Conformity Manufacturer: Soteria Biotech Co., Ltd.

12F, No. 365, Fude 3rd Rd., Xizhi Dist., New Taipei City 22151 Taiwan SRN: TW-MF-000013120

European Authorised MT Promedt Consulting GmbH Representative: Ernst-Heckel-Straße 7 66386 St. Ingbert, Germany SRN: DE-AR-000000085 Product Name: Soteria One Point Zero (version: 1.0) Internal software version: Soteria one point zero-20170721143733 Basic UDI-DI: 471988033001FV GMDN Code: 41670 Radiological picture archiving and communication system (PACS) software UMDNS Code: 16247 Picture archiving and communication systems (PACS), Radiology Intended purpose: Soteria One Point Zero is a cloud based standalone software platform designed to display and analyze head and neck medical images Classification: Class IIa, Rule 11 per Annex VIII of Regulation 2017/745 Conformity Assessment Annex IX Chapter I and III of Regulation 2017/745 Route: We herewith declare that the above mentioned product meet the provisions of Medical Device Regulation 2017/745. All supporting documentation is retained under the premises of the manufacturer. Common Specifications N/A Used:

Doc#: OPZ-021-05 Page 2 of 2

Notified Body: TÜV SÜD Product Service GmbH Zertifizierstellen, Ridlerstraße 65, 80339 MÜNCHEN Germany

NB Identification number: CE0123 Certificate No.: G1 102623 0002 Rev. 00 Issue Date: 2020-01-07 Expiry Date: 2024-05-26 The aforementioned declaration of conformity is exclusively under the responsibility of the CEO of Soteria Biotech Co., Ltd. __________________________ __________________________________ Place, Date Hung-Ta Hsiao, CEO

VLV TRANS SIA "Saulaines", Bilskas pag., Smiltenes nov., LV-4706

Latvia [email protected]

+371 27873909

THE DATE OF THE DOCUMENT SHALL CORRESPOND TO THE DATE OF THE LAST SECURED ELECTRONIC SIGNATURE AND THE ATTACHED TIME STAMP

No. VLV-2025-051

Karmen Joller Minister of Social Affairs Ministry of Social Affairs Suur‑Ameerika 1 10122 Tallinn, Estonia [email protected] [email protected]

Proposal for Implementing Soteria One Point Zero in Estonian Healthcare

Dear Minister Joller,

On behalf of VLV Trans SIA, a company dedicated to advancing innovative healthcare solutions, I am pleased to present Soteria One Point Zero, an innovative AI-powered platform designed to revolutionize the diagnosis and management of obstructive sleep apnea (OSA). As a partner of Soteria Biotech Co., Ltd., VLV Trans SIA is committed to introducing cutting-edge medical technologies to Estonia, the goal of enhancing healthcare outcomes and operational efficiency.

Key Advantages of Soteria One Point Zero:

 Efficiency: Processes patient CT/CBCT scans in under two hours, offering precise, actionable reports.

 Diagnostic Precision: Achieves up to 91% accuracy in OSA diagnostics, as validated through extensive testing on over 6,800 cases.

 Convenience: Minimizes reliance on overnight sleep studies, enabling rapid analysis through Computational Fluid Dynamics (CFD) simulations.

 Accessibility: Removes barriers to care for underserved patient groups, including children, elderly, and high-risk individuals.

 Cost-Effectiveness: Reduces administrative workload, saving time and resources for healthcare providers.

Proven Performance

Soteria One Point Zero is CE-certified and adheres to ISO 13485 standards, ensuring safety and efficacy. The software uses advanced algorithms to simulate airflow dynamics, pinpointing

obstructions that are difficult to detect through conventional methods. This enhances clinical decision-making and accelerates treatment planning.

The platform has demonstrated exceptional results in diverse healthcare settings, including partnerships with leading hospitals in Taiwan, Singapore, and the Netherlands. Its adaptability to different clinical environments makes it a valuable asset for Estonia’s healthcare system.

Supporting Documentation

To further support our proposal, we have included additional documentation with this letter. These materials provide detailed insights into the technical specifications, clinical performance, and certifications of Soteria One Point Zero. We invite you to review these documents for a deeper understanding of the software’s potential impact.

Trial and Cost Structure

To showcase the transformative potential of our software, we offer a one-month free trial for Estonian medical institutions. During this trial, the platform can be utilized without limitations to experience its benefits firsthand.

Following the trial period, the cost for using the software in a specific field is equivalent to the daily cost of serving one patient, according to the medical institution’s cost table. Any revenue generated beyond this threshold remains with the medical institution, creating an incentive for increased usage and integration into routine care. While this standard pricing model serves as a guideline, the exact terms can be discussed and agreed upon during negotiations to ensure mutual satisfaction.

Collaboration Opportunity

If the Ministry is interested in exploring this innovative solution further, we propose an online meeting via Zoom with our product inventor and CEO of Soteria Biotech Co., Ltd, Mr. Hung Ta Hsiao, to discuss implementation strategies and answer any questions.

Soteria One Point Zero offers a unique opportunity to revolutionize healthcare delivery in Estonia by addressing critical challenges in OSA diagnosis and treatment. We are confident that this collaboration can significantly improve patient outcomes while optimizing healthcare resources.

Thank you for considering our proposal. We look forward to the possibility of working together to advance healthcare standards in Estonia.

Yours sincerely, Vilnis Lībietis Member of the Board

THE DOCUMENT HAS BEEN SIGNED WITH A SECURE ELECTRONIC SIGNATURE AND CONTAINS A TIME STAMP.

Page 1 of 3

Manufacturer’s Declaration

in relation to Regulation (EU) 2023/607 amending Regulations (EU) 2017/745 and (EU) 2017/746 as

regards the transitional provisions for certain medical devices and in vitro diagnostic medical devices, in

particular with respect to

 the validity of certificates issued under Council Directive 90/385/EEC on Active Implantable Medical

Devices (AIMDD) or Council Directive 93/42/EEC on Medical Devices (MDD) (Directive Certificates)

and/or1

 the compliance of the devices and us as their manufacturer with the conditions for the continued

placing on the market and putting into service

Manufacturer name Soteria Biotech Co., Ltd.

Manufacturer address and contact details

12F, No.365 Fude 3rd Rd., Xizhi Dist.,

New Taipei City 22151

Taiwan

Single Registration Number (SRN) (if available) TW-MF-000013120

Authorised Representative name (if applicable) MT Promedt Consulting GmbH

Authorised Representative address and contact details

Ernst-Heckel-Straße 7

66386 St. Ingbert,

Germany

Single Registration Number (SRN) (if available) DE-AR-000000085

Notified body name (if applicable) TÜ V SÜ D Product Service GmbH

□ See attached schedule

Notified body number (if applicable) CE0123

□ See attached schedule

Directive Certificate number(s) to which this confirmation is made (if applicable)

G1 102623 0002 Rev. 00

□ See attached schedule

Original expiry date as indicated on the Directive Certificate prior to the extension of the validity (if applicable)

26 May 2024

□ See attached schedule

End date of extended validity/transition period 31 December 2028

□ See attached schedule

1 The first condition is not applicable in case of devices for which the conformity assessment procedure pursuant to MDD did not

require the involvement of a notified body, for which the declaration of conformity was drawn up prior to 26 May 2021 and for which the conformity assessment procedure pursuant to this Regulation requires the involvement of a notified body.

Page 2 of 3

We, as the manufacturer declare under our sole responsibility:

 for the above listed Directive Certificate (or see attached schedule, if multiple certificates) the

conditions for the legal extension of validity as required in Article 120.2 of the MDR are met and/or2

 the listed device(s) in the attached schedule and we as their manufacturer are in compliance with

the conditions listed in Article 120.3c of the MDR for continued placing on the market and putting

into service,

namely by fulfilling the following conditions:

 Directive Certificate(s) as listed above or in the attached schedule

 Directive Certificate(s) covering the listed device(s) was/were issued after 25 May 2017, was/were

valid on 26 May 2021 and have not been withdrawn afterwards.

Expired/expires after 20 March 2023:

Formal application(s) to the notified body in accordance with Section 4.3, first subpara-

graph of Annex VII MDR for conformity assessment has/have been made or will be

made/submitted by us to a notified body no later than 26 May 2024 for the device(s) listed

in the attached schedule or its/their substitute(s) and signed written agreement(s) is/will be

in place in accordance with Section 4.3, second subparagraph of Annex VII MDR before

26 September 2024.

 Quality Management System (QMS)

A QMS in accordance with Article 10(9) MDR will be put in place by no later than 26 May 2024.

 Device(s) as listed in the attached schedule

 The device(s) continue to comply with the AIMDD or MDD.

 There are no significant changes in the design and intended purpose.

 The device(s) do not present an unacceptable risk to health or safety of patients, users or other

persons, or to other aspects of the protection of public health.

Signed for and on behalf of the manufacturer:

Soteria Biotech Co., Ltd.

New Taipei City, Taiwan January 4, 2024

Hung Ta Hsiao, CEO

[email protected]

2 The first condition is not applicable in case of devices for which the conformity assessment procedure pursuant to MDD did not

require the involvement of a notified body, for which the declaration of conformity was drawn up prior to 26 May 2021 and for which the conformity assessment procedure pursuant to this Regulation requires the involvement of a notified body

Page 3 of 3

Schedule of Devices The above Manufacturer’s Declaration is valid for the following devices:

Identification of the device(s)3 (e.g., device name, family/group name device model or catalogue number)

Directive Certificate number(s) to which this confirmation is made (if applicable)

Original expiry date as indicated on the Directive Certificate (s) prior to the extension of the validity (if applicable)

Notified Body name and number that issued the Directive Certificate (if applicable)

Notified Body name and number where the MDR application was lodged/contract signed (if applicable)

End date of extended validity / transition period

Substitute Device(s) (if applicable)

Soteria One Point Zero

G1 102623 0002 Rev. 00

26 May 2024 TÜ V SÜ D Product Service GmbH CE0123

31 December 2028 N/A

3 for devices with AIMDD/MDD certificate(s) the identification should be as in the certificate, and only if the certificate has a generic scope it should be as defined

above)

Soteria Biotech Co., Ltd.
 2024.07.22

CT / CBCT
 Imaging Requirements for all Customers

Suitable CBCT Examples *or any available CT is suitable

Viso® family of premium CBCT imaging units Viso® family of premium CBCT imaging units Panoramic X-ray system eco-x

Standard CT Image (Hospitals)

• Head and Neck CT


• At least the area in the red box


• From the circled nasal cavity and down to the hyoid bone

Pixel size: 0.0625cm x 0.0625cm x 0.0625cm or better

Standard CBCT Image (Dental) Pixel size: 0.0625cm x 0.0625cm x 0.0625cm or better Shooting range: FOV > 10cm x 10cm or greater

• Head and Neck CT


• At least the area in the red box


• From above the end of the circled hard palate and down to the hyoid bone

CT / CBCT Shooting Method

1. Before the shooting, the patient should: • lie down (CT) or sit/stand (CBCT), • be in centric occlusion, • completely exhale all air from chest

and abdomen.


2. During the shooting, the patient should: • keep holding exhaled state, • strictly avoid swallowing.

*contrast agent isn’t required

1

Revision History

Revision No. Revision Description Publication Date Author

01 Initial release 2019/05/02 Hsiao

2

Contents

1. Introduction ............................................................................................................................................. 3 2. Purpose of the report ............................................................................................................................ 3 3. Product description ................................................................................................................................ 3 4. Intended use ........................................................................................................................................... 3 5. Indication for use .................................................................................................................................... 3 6. Method ..................................................................................................................................................... 4 7. Selection and exclusion of Subjects ................................................................................................... 6 8. The specification of CBCT scanner .................................................................................................... 7 9. Assessment of Efficacy ......................................................................................................................... 7 10. Statistics .................................................................................................................................................. 8 11. Assessment of Saftey ........................................................................................................................... 8 12. Results ..................................................................................................................................................... 9 13. Conclusion ............................................................................................................................................ 11 14. Reference .............................................................................................................................................. 11

3

1. Introduction

This document describes the performance of Soteria One Point Zero (version no.: 1.0). This

report is prepared with the consideration of following references:

– Software as a Medical Device (SAMD): Clinical Evaluation – Guidance for Industry and

Food and Drug Administration Staff (December 8, 2017);

– Clinical Evaluation: A Guide for Manufacturers and Notified Bodies Under Directives

93/42/EEC and 90/385/ECC (MEDDEV 2.7/1 revision 4)

2. Purpose of the report

This performance report is a late-development-stage report. The goal is to establishing a

baseline of the product performance and as an evidence that Soteria One Point Zero can be

used safely and efficiently.

3. Product description

Soteria One Point Zero is a cloud based stand-alone software platform designed to display and

analyze head and neck medical images. The software utilizes automatic airway detection

algorithm / manual airway selection, 3-D model reconstruction, and computational fluid

dynamics (CFD) to simulate and calculate for flow characteristics include flow velocity and

pressure within the airway. These information can be used to determine if the airway is

obstructed and at which location.

4. Intended use

Soteria One Point Zero is a cloud based stand-alone software platform designed to display and

analyze head and neck medical images.

5. Indication for use

Soteria One Point Zero is a cloud based stand-alone software platform designed to display and

analyze head and neck medical images (DICOM compatible images) for airway flow

4

characteristics including flow velocity and pressure within the airway. These results are intended

to be used along with patient’s clinical history, as well as physicians’ professional judgment.

These information are not for immediate diagnostic nor treatment use, they are used to

determine if the airway is obstructed and at which location.

6. Method

This study included 200 subjects from different sites, sites include Chang-Gang Memorial

Hospital (Taoyuan, Taiwan, with 89 subjects), China Medical University Hospital (Taichung,

Taiwan, with 71 subjects), and Kaohsiung Medical University Hospital (Kaohsiung, Taiwan,

with 40 subjects). Subjects were grouped into control group (subject without obstructive

sleep apnea, OSA) and OSA group by the diagnosis.

Obstructive sleep apnea is a common sleep disorder of adults and children, it is

characterized by repetitive transient reversible upper airway obstructions during sleep.

In general, patient with OSA has narrower upper airway compare with subject without

OSA. According to American Academy of Sleep Medicine (AASM), OSA is diagnosed by

using the polysomnography (PSG) result plus clinical symptoms, this is currently the gold

standard for OSA diagnosis. The result of PSG is presented as apnea hypopnea index or

AHI and it is defined as number of apnea (complete cessation of airflow for at least 10

seconds) or hypopnea (airflow decreases by 50% for 10 seconds or decreases by 30% if

there is an associated decrease in the oxygen saturation or an arousal from sleep) in an

hour. The severity of OSA is classified by AHI where AHI < 5 per hour is considered as

normal or minimal, AHI ≥ 5, but < 15 per hour is mild, AHI ≥ 15, but < 30 per hour is

moderate, and AHI ≥ 30 per hour is considered as severe.

The study procedure flow chart is shown in Figure 1, subject first get diagnosis by the first

clinician with gold standard method. The same subject then takes the Cone Beam

Computed Tomography (CBCT) scan, the scanned image is analyzed by the software and

the report is presented to the second clinician and the clinician determines if subject’s

5

airway is obstructed. Finally the results from the first and the second clinicians are

compared.

The specification of CBCT scan is shown in Figure 2.

Figure 1. Study flow chart

Figure 2. The region of CBCT scan. The red rectangle region is around 7-9cm; Pixel size: 0.06cm * 0.06cm * 0.06cm;

6

The criteria for determine subject with or without airway obstruction is shown in Table 1 below:

Table 1. Criteria for determine subject with or without airway obstruction

Software report Condition

The color distribution shows a uniform color which means minimal pressure or velocity change. The interpretation is the air flow through this section of the airway smoothly without obstruction Condition = Without obstruction

The color distribution showed color difference at the section boxed in red which means pressure difference at that location. The interpretation is the air flow through that section with some obstruction causing change of the pressure and velocity. Condition = Airway obstruction

The color distribution showed a sharp color change from green to dark blue at the section boxed in red which means a large pressure difference at that location. The interpretation is the air flow through this section of the airway with obstruction causing sharp change of the pressure and velocity. Condition = Airway obstruction

7. Selection and Exclusion of Subjects

Selection Criteria: 1. Male or female subjects age equal to or above 20, equal to or below 65 2. Polysomnography (PSG) study, with study time for at least 5 hours or above 3. BMI equal to or less than 30

7

Exclusion Criteria: 1. Severe uncontrollable cardiovascular, neurological or metabolic diseases 2. Cannot take x-ray examination 3. Pregnant or possible pregnant women 4. Central sleep apnea 5. Subject had head and neck surgery (may affect the scanning result)

Discontinuation of study: If subject is not feeling well during any part of study, subject can decide to terminate the study

8. The specification of CBCT scanner

The specification of CBCT scanner used is shown in Table 2.

Table 2. Specification of CBCT scanner

X-ray Tube

Voltage 60~120kV

Current 1~20mA

Focal Spot Size 0.3~0.5mm Signal Gray Scale 12~16bit Image Detector <40cm×40cm Source to Detector Distance >50cm Source to Patient Distance >48cm Field Of View(FOV) (6~23)×(6~23) Scan Time(s) <30 Image Acquisition 180°~360° Reconstruction Time(s) <120 DICOM File Size 50M~2G Voxel Size <0.4mm×0.4mm×0.4mm

9. Assessment of Efficacy

This study is to use the criteria shown in Table 1 to determine if the airway of the subject is

obstructed. The comparison is done by compare the clinician determined result using the

8

software vs. gold standard diagnostic result. If the accuracy of the result determined by

clinician using the software compared with gold standard diagnostic result is 70% or above, it

is considered as effective.

10. Statistics

The validity of the index test is measured with sensitivity and specificity shown in two-by two

table below

Table 3. 2x2 table

Cell “a” we enter those who have positive results for the index test and the gold standard test

also tested positive. These are the true positive.

Cell “b” we enter those who have positive results for the index test but do not test positive

according to the gold standard test. These are false positive.

Cell “c” we enter those who have negative results for the index test but positive results on the

gold standard test. These are false negative.

Cell “d” we enter those who have negative results for the index test and the gold standard test

also test negative. These are the true negative.

Sensitivity is the ability of a test to correctly identify an individual as positive, it is also called

true positive rate. Sensitivity= a/(a+c) which is probability of being tested positive when disease

is present. Specificity is the ability of a test to correctly identify an individual as negative, it is

also called true negative rate. Specificity= d/(b+d) which is probability of being tested negative

when disease is absent. Accuracy is the sum of the true positive and the true negative among

all tested. Accuracy = a+d/(a+b+c+d).

9

11. Assessment of Safety

The specification of CBCT is shown in table 2. In accordance with the scanning area (Figure 2), the maximum dosage per scan is approximately 120μSv. According to the international commission on radiological protection, ICRP, the limitation of radiation dosage for an adult is 5 mSv (5000μSv ) per year. Therefore, as long as subject does not expose to radiation exceeding the mentioned limitation, there should be no harm from the radiation.

12. Results

Figure 3 shows the air flow pressure color distribution of subject’s airway with AHI of 7.9, 13.5,

and 81. As AHI gets higher, the pressure difference (color difference) is also higher. It can clearly

seen the largest pressure difference is at the location where the airway is narrowest (boxed in

red).

Figure 3. The airflow pressure color distribution of subjects with different AHI

Table 4 shows the relative physiological data, subjects are divided into Group 1 (AHI<5) 63

subjects, Group 2 (5≦AHI<15) 8 subjects, Group 3 (15≦AHI<30) 42 subjects, and Group 4

(AHI≧30) 87 subjects.

10

Table 4. The relative physiological data of subjects

Group1(AHI<5) Group2(5<AHI<15) Group3(15<AHI<30) Group4(AHI>30)

case 63 8 42 87

Sex Male 42 6 33 58

Female 21 2 9 29

Age 49.4±12.2 39.4 ±12.7 49.3 ±12.2 54.3 ±12.5

AHI 1.9 ±1.5

8.9 ±3.3 20.7±4.1 55.2 ± 15.7

Table 5 shows that doctor use the report from Soteria One Point Zero to evaluate whether

subject’s airway has obstruction and got sensitivity, specificity, and accuracy above 80%.

Table 5. Compare PSG (gold standard) with doctor use the report from Soteria One Point Zero

PSG Diagnosis

(AHI≧5 is positive)

Positive

(AHI≧5)

Negative

(AHI<5) Row Total

Doctor’s evaluation use report from

Soteria One Point Zero

Positive 121 8 129

Negative 16 55 71

Column Total 137 63 200

Sensitivity=121/137=0.88

Specificity=55/63=0.87

Accuracy =（121+55）/200=0.88

11

13. Conclusion

This study is a blind test with OSA patient and we conclude the report from Soteria One Point

Zero is helpful to doctors to determine if subjects has airway obstruction and with accuracy of

88%.

14. Reference

Caples, S. M., T. Kara, et al. (2005). "Cardiopulmonary consequences of obstructive sleep apnea." Semin

Respir Crit Care Med 26(1): 25-32.

El-Ad, B. and P. Lavie (2005). "Effect of sleep apnea on cognition and mood." Int Rev Psychiatry 17(4):

277-82.

Fogel, R. B., A. Malhotra, et al. (2004). "Sleep. 2: pathophysiology of obstructive sleep apnoea/hypopnoea

syndrome." Thorax 59(2): 159-63.

Katz, I. M., B. M. Davis, et al. (1999). "A numerical study of particle motion within the human larynx and

trachea." J Aerosol Sci 30: 11.

Keyhani, K., P. W. Scherer, et al. (1995). "Numerical simulation of airflow in the human nasal cavity." J

Biomech Eng 117(4): 429-41.

Lam, B., C. G. Ooi, et al. (2004). "Computed tomographic evaluation of the role of craniofacial and upper

airway morphology in obstructive sleep apnea in Chinese." Respir Med 98(4): 301-7.

Leiter, J. C. (1996). "Upper airway shape: Is it important in the pathogenesis of obstructive sleep apnea?" 、

Am J Respir Crit Care Med 153(3): 894-8.

Li, W. I., M. Perzl, et al. (1996). "Aerodynamics and aerosol particle deaggregation phenomena in model

oral-pharyngeal cavities." J Aerosol Sci 27: 18.

Liao, Y. F., C. S. Huang, et al. (2003). "The utility of cephalometry with the Muller maneuver in evaluating

the upper airway and its surrounding structures in Chinese patients with sleep-disordered breathing."

Laryngoscope 113(4): 614-9.

Malhotra, A., Y. Huang, et al. (2002). "The male predisposition to pharyngeal collapse: importance of airway

length." Am J Respir Crit Care Med 166(10): 1388-95.

Pae, E. K., A. A. Lowe, et al. (1997). "A role of pharyngeal length in obstructive sleep apnea patients." Am J

Orthod Dentofacial Orthop 111(1): 12-7.

Polo, O. J., M. Tafti, et al. (1991). "Why don't all heavy snorers have obstructive sleep apnea?" Am Rev

Respir Dis 143(6): 1288-93.

Remmers, J. E., W. J. deGroot, et al. (1978). "Pathogenesis of upper airway occlusion during sleep." J Appl

Physiol 44(6): 931-8.

Schwab, R. J. (1996). "Properties of tissues surrounding the upper airway." Sleep 19(10 Suppl): S170-4.

Schwab, R. J., W. B. Gefter, et al. (1993). "Dynamic upper airway imaging during awake respiration in

normal subjects and patients with sleep disordered breathing." Am Rev Respir Dis 148(5): 1385-400.

Schwab, R. J., W. B. Gefter, et al. (1993). "Dynamic imaging of the upper airway during respiration in

normal subjects." J Appl Physiol 74(4): 1504-14.

Stierer, T. and N. M. Punjabi (2005). "Demographics and diagnosis of obstructive sleep apnea." Anesthesiol

12

Clin North America 23(3): 405-20, v.

Subramaniam, R. P., R. B. Richardson, et al. (1998). "Computational fluid dynamics simulations of

inspiratory airflow in the human nose and nasopharynx." Inhalation Toxicology 10(5): 30.

Vos, W., J. De Backer, et al. (2007). "Correlation between severity of sleep apnea and upper airway

morphology based on advanced anatomical and functional imaging." J Biomech 40(10): 2207-13.

Useful Q&A on Soteria AI platform:

Q: How many OSA cases are being operated by Soteria platform for AI system training at the moment [September 2022]? Is there any post-operative feedback optimization training?

A: Soteria AI system training is currently based on a total of 6872 cases (Taiwan 6840, the UK 20, the Netherlands 12). When the amount of cases had reached 200, the OSA Report accuracy rate reached 88%. Then Soteria successfully passed the TFDA and CE certification. The missing percentage is mainly mild OSA cases, since it’s challenging to collect data from patients with mild symptoms. The accuracy rate of post-treatment optimization training is currently 90% (published at the 2018 Asian Sleep Conference).

Q: How many patients have been successfully assisted by Soteria AI platform at the moment? How did it become possible to predict post-treatment improvements with such high efficacy?

A: Soteria has registered 6840 patients’ upper airways improvement, based on provided clinical feedback in Taiwan. Initially, for the purpose of TFDA and CE certification, Soteria has participated in 200+ PSG diagnostics in sleep centers. PSG complete output data was fed to the AI program together with pre-treatment CT images of the same patients to detect corresponding information. Post-treatment CT images of the same patients followed up feeding the AI program to further ‘teach’ it to recognize results and predict them with just the input of CT images only. Since then, by fulfilling the contract, Kaohsiung Medical University Hospital (Kaohsiung city, Taiwan) is continuing to provide pre- and post-treatment detailed clinical feedback to Soteria to refine and advance our medical software. Also, this key corresponding information uncovered by the system can’t be seen by a doctor on a CT scan, it can only be simulated and calculated by the AI and offered as visual colorful image of airflow pressure and velocity. Basically, now a doctor only needs to input CT image, and receive complete+ gold standard diagnostic result without asking a patient to stay in a sleep center.

Q: Will the treatment methods/ standard definitions/ rehabilitation definitions be different in different countries and regions? Will the above definitions or methods have a certain influence on the judgment capability of Soteria AI platform?

A: For OSA, these treatment methods/ standard definitions/ rehabilitation definitions in different countries are all guided by the criteria of the American Sleep Society and the European Sleep Society, so there is no difference.

At present, there are no differences detected among different human races in the OSA test results conducted in the Netherlands and the UK. Due to the uniqueness of Soteria AI+CFD technology, the analysis result is based on breathing pressure and breathing speed, not on the conventional respiratory tract volume and cross-sectional area. Thus, belonging to a different human race creates zero influence on Soteria OSA evaluation capability. In other words, the software can be efficiently applicable for any patient.

Q: In addition to the FDA and CE certification, does the AI platform have any exclusive doctor/university certification?

A: Currently, thoracologists, otolaryngologists, and neurologists in Taiwan are all Soteria platform users and have published medical journal papers for certification.

Q: Are there any requirements for CT scanning while standing/lying down, or leaning forward/backward/sideways, etc.?

A: When scanning while lying down, patient should lie flat horizontally. When scanning while standing, patient should stand up straight. Patient shouldn’t move/tilt his/her head. Patient should completely exhale, then radiologist starts scanning.

Q: How many times CT scans are usually performed for different recommended solutions? If the same case requires multiple scans and analyses, is the cost being also charged for a single case?

A: There are mainly these 3 different treatment plans: 1.Respirator (CPAP): only needs to scan once —> single charge. 2.Oral Sleep Device: two shots before treatment —> one case —> single charge. 3.Orthodontic/Laryngeal Surgery (MMA): once before the operation, once after the operation —> the cost of the case is charged twice.

Q: What level of details can AI platform possibly provide with the guidelines/risk warning? For example, will it only be “Surgery is required”, or something like "Tracheal dilation surgery is required (the cut part is marked in the red area of the image), pay attention to the adhesion of the lower part" and the like?

A: Soteria report mainly provides the location of the respiratory tract obstruction, the severity of the obstruction, and the cause of the blockage. The detailed guidelines (as in the example above) will not be recommended by AI platform. The physician proposes the final solution with the help of all information provided in the report.

Q: Is it convenient to provide a test account to understand the entire platform operation process?

A: Sure. Just click https://lydata.com.tw/lydata_soteriabio/admin/index.html and give us a call to get your login and password.

Q: Is it convenient to provide instructions for making standard CT images for reference?

A: Sure. Please refer to the corresponding file.

Q: Will there be some public information, brochures or posters that can be provided to the hospitals to exhibit/post/introduce to patients?

A: Yes, Soteria can provide you with digital copies for printing.

Q: If patient needs to use oral sleep device, does Soteria also provide them, or can the local hospitals make these braces themselves?

A: Soteria can also provide oral sleep devices, which in turn require intraoral scan files and CT files, and additional business negotiations and a whole new business setup. Or the local hospitals can make their own braces.

Q: If patient has undergone surgery and the trachea is in unnatural state, does the postoperative situation apply to this AI evaluation?

A: Yes, it is applicable. AI platform can also provide an analysis report on the degree of narrowing of the respiratory tract, and the rest of the information in the same way as for a natural state.

Q: Will the AI platform store records of patient data? If so, where will they be stored (server location)?

A: No, Soteria doesn’t store records of patient data. In accordance with medical regulations and ISO13485 regulations, the uploaded image files will first be stored in a cloud company that meets the information security specifications. European data is stored in Amsterdam (Netherlands), Taiwanese data is stored in Hsinchu (Taiwan).

Patient OSA analysis data (Report) will be automatically deleted in two days after it has been downloaded by the hospital.

Q: How long are the Reports stored on the platform?

A: In accordance with medical regulations and ISO13485 regulations, patient OSA analysis data (Report) will be automatically deleted in two days after it has been downloaded by the hospital. As long as there's business with the hospital, hospital can choose to store used reports on the platform. However, based on Soteria experience, hospitals rather prefer to keep downloaded reports (PDF format) in the hospital. If business with the hospital is terminated, hospital will still be granted with some time to save all paid reports as PDF files, and then all relevant data will be deleted.

Some straight-forward FAQs from doctors’ perspective:


 Q1: Sounds like you are trying to load me with something new, that requires to get studied, get used to... And I’m already way beyond busy!

A1: On the contrary, Soteria is designed to UNLOAD your hectic schedule. Expert’s time and direct attention to patients are most precious. The rest of the routine that required 1-3 days per visitor, now can be done in 1-2 hours. Just drag and drop CT scans to our online platform, and receive OSA analysis Report.
 What Doctor DOES see on Soteria platform: it processes uploaded CT images and issues a Report with the same but colored CT images and detailed crucial data, exceeding PSG’s.
 What Doctor DOES NOT see on Soteria platform (it’s background functioning): our system is “trained” to simulate the work of upper respiratory tract during sleep WITHOUT the need for the Patient to actually sleep in the laboratory.
 The complete process is elegantly simplified for your ultimate convenience!
 In other words, WHY should more and more Patients continue sleeping in laboratories, when thousands and thousands of other Patients with different degrees of OSA complexity have already slept before them?
 WHY spending 1-3 days on one Patient, when for the same retail value and with great accuracy you can examine him/her in an hour or two? And our Report will show the result of simulation, how Patient's upper airways behaved while “sleeping”.

Q2: Are you trying to replace Doctors with AI? :)

A2: No, it’s illegal, and/or must be so. Doctor’s expertise is INDISPENSABLE! The software does NOT make decisions INSTEAD of a doctor, it only analyzes big data by just being a "smart calculator”.


The software removes 80% of work routine: collecting PSG/other measurements, data entry, simulation, calculation, analysis; and outputs visually intuitive (also patient- friendly) data.
 20% of the work can only be done by an experienced doctor, relying on the Report to choose the best method of treatment. And these 20% are most important.

Q3: We’ve (just) invested a ton in the best sleep center equipment! And now you want us to replace Gold Standard PSG with AI?

A3: At that moment it surely was your best investment in business growth and service quality. Today, why not to have both at your disposal? You already have a certain number of beds in your sleep laboratory. How about “installing” more “beds”, even more “sleep centers” with a small fraction of their cost?
 These virtual “beds” DON’T require: space in your clinic, more equipment and consumables, maintenance, new training, qualified personnel, extra services for Patients during the whole procedure, extra attention from Doctors, etc.
 INSTEAD, these virtual “beds” provide you with precise certified analyses AND attract largely underexamined Patients who did not want to come previously, because:
 --clinics don’t offer quick diagnosis,
 --it takes long waiting time to finally get examined in the laboratory,
 --they’re busy and can't set aside as much as 1-3 days for this procedure,
 --their OSA has unnoticeable or light form, which they believe is unworthy of 1-3 days spent in a hospital,
 --they feel it’s too complex and morally uncomfortable,
 --they hesitate to visit any medical institution due to fear of Covid-19 or other contagious diseases,
 --they are children (including: they’re just too small; with lagging development; with hysterical behavior; underachieving pupils with ADHD; who require orthodontic treatment), 
 --they are old people, disabled (with caretakers), weak (Patients who need to identify whether OSA is one of the root reasons why they’re sick for a long time, don’t recover well, recover poorly after surgery).
 All these categories of Patients are not even included in the clinical OSA studies, and thus this market is largely underpenetrated.
 So, now you see how you can rethink your service quality and business EFFICIENCY with Soteria’s improved approach. Our software helps to EXPAND your business without inserting a stick into the wheel of what has already been established by your esteemed clinic.

Q4: Why do we need your AI system if Pulmonologist can see where the narrowing is on the CT scan, and whether surgery is needed or not?

A4: Of course, complex and trouble-free cases are always obvious for Pulmonologist, but all in between – not so much.
 Our system is especially good when it’s difficult for а Doctor to determine by eye whether there is an OSA. That’s why Soteria has created visual Reports: air flows are

highlighted in colors, how they move, where they get stuck, pressure, velocity, etc. – all during Patient’s simulated "sleep”.
 Doctor can’t possibly SEE how Patient’s upper airways behave when he snores and suffocates. AI can. This process can only be “SEEN” with Computational Fluid Dynamics (CFD) simulation; with PSG/else – it’s just a bunch of numbers (with some inconsistency).
 In 2016–2017, the medical software Developer, together with leading Pulmonologists in Taiwan, conducted over 200 comprehensive measurements and studies in the sleep centers of Taiwan's leading hospitals to collect high quality big data from a wide variety of OSA cases, and "teach" AI to "see" all changes in the Patients’ breathing patterns in their sleep.
 Nowadays, the world's first patented and CE certified system is operating thousands of such cases, constantly self-learning and enhancing its Reports’ precision beyond current 88%–91%.

Q5: Can you provide contacts of Doctors who have already implemented your system? We’d like to personally hear their feedback.

A5: Currently, there’re over 15 hospitals, medical centers and clinics in Taiwan, Singapore, the Netherlands, and the UK, that are successfully using Soteria platform and satisfied with OSA Reports.
 Please send us a message to [email protected] or give us a call at… (WhatsApp, Telegram), and we’ll arrange a conversation with an available English-speaking doctor without breaching existing NDAs (Non-Disclosure Agreements).

Doc#: OPZ-021-05 Page 1 of 2

EC Declaration of Conformity Manufacturer: Soteria Biotech Co., Ltd.

12F, No. 365, Fude 3rd Rd., Xizhi Dist., New Taipei City 22151 Taiwan SRN: TW-MF-000013120

European Authorised MT Promedt Consulting GmbH Representative: Ernst-Heckel-Straße 7 66386 St. Ingbert, Germany SRN: DE-AR-000000085 Product Name: Soteria One Point Zero (version: 1.0) Internal software version: Soteria one point zero-20170721143733 Basic UDI-DI: 471988033001FV GMDN Code: 41670 Radiological picture archiving and communication system (PACS) software UMDNS Code: 16247 Picture archiving and communication systems (PACS), Radiology Intended purpose: Soteria One Point Zero is a cloud based standalone software platform designed to display and analyze head and neck medical images Classification: Class IIa, Rule 11 per Annex VIII of Regulation 2017/745 Conformity Assessment Annex IX Chapter I and III of Regulation 2017/745 Route: We herewith declare that the above mentioned product meet the provisions of Medical Device Regulation 2017/745. All supporting documentation is retained under the premises of the manufacturer. Common Specifications N/A Used:

Doc#: OPZ-021-05 Page 2 of 2

Notified Body: TÜV SÜD Product Service GmbH Zertifizierstellen, Ridlerstraße 65, 80339 MÜNCHEN Germany

NB Identification number: CE0123 Certificate No.: G1 102623 0002 Rev. 00 Issue Date: 2020-01-07 Expiry Date: 2024-05-26 The aforementioned declaration of conformity is exclusively under the responsibility of the CEO of Soteria Biotech Co., Ltd. __________________________ __________________________________ Place, Date Hung-Ta Hsiao, CEO

VLV TRANS SIA "Saulaines", Bilskas pag., Smiltenes nov., LV-4706

Latvia [email protected]

+371 27873909

THE DATE OF THE DOCUMENT SHALL CORRESPOND TO THE DATE OF THE LAST SECURED ELECTRONIC SIGNATURE AND THE ATTACHED TIME STAMP

No. VLV-2025-051

Karmen Joller Minister of Social Affairs Ministry of Social Affairs Suur‑Ameerika 1 10122 Tallinn, Estonia [email protected] [email protected]

Proposal for Implementing Soteria One Point Zero in Estonian Healthcare

Dear Minister Joller,

On behalf of VLV Trans SIA, a company dedicated to advancing innovative healthcare solutions, I am pleased to present Soteria One Point Zero, an innovative AI-powered platform designed to revolutionize the diagnosis and management of obstructive sleep apnea (OSA). As a partner of Soteria Biotech Co., Ltd., VLV Trans SIA is committed to introducing cutting-edge medical technologies to Estonia, the goal of enhancing healthcare outcomes and operational efficiency.

Key Advantages of Soteria One Point Zero:

 Efficiency: Processes patient CT/CBCT scans in under two hours, offering precise, actionable reports.

 Diagnostic Precision: Achieves up to 91% accuracy in OSA diagnostics, as validated through extensive testing on over 6,800 cases.

 Convenience: Minimizes reliance on overnight sleep studies, enabling rapid analysis through Computational Fluid Dynamics (CFD) simulations.

 Accessibility: Removes barriers to care for underserved patient groups, including children, elderly, and high-risk individuals.

 Cost-Effectiveness: Reduces administrative workload, saving time and resources for healthcare providers.

Proven Performance

Soteria One Point Zero is CE-certified and adheres to ISO 13485 standards, ensuring safety and efficacy. The software uses advanced algorithms to simulate airflow dynamics, pinpointing

obstructions that are difficult to detect through conventional methods. This enhances clinical decision-making and accelerates treatment planning.

The platform has demonstrated exceptional results in diverse healthcare settings, including partnerships with leading hospitals in Taiwan, Singapore, and the Netherlands. Its adaptability to different clinical environments makes it a valuable asset for Estonia’s healthcare system.

Supporting Documentation

To further support our proposal, we have included additional documentation with this letter. These materials provide detailed insights into the technical specifications, clinical performance, and certifications of Soteria One Point Zero. We invite you to review these documents for a deeper understanding of the software’s potential impact.

Trial and Cost Structure

To showcase the transformative potential of our software, we offer a one-month free trial for Estonian medical institutions. During this trial, the platform can be utilized without limitations to experience its benefits firsthand.

Following the trial period, the cost for using the software in a specific field is equivalent to the daily cost of serving one patient, according to the medical institution’s cost table. Any revenue generated beyond this threshold remains with the medical institution, creating an incentive for increased usage and integration into routine care. While this standard pricing model serves as a guideline, the exact terms can be discussed and agreed upon during negotiations to ensure mutual satisfaction.

Collaboration Opportunity

If the Ministry is interested in exploring this innovative solution further, we propose an online meeting via Zoom with our product inventor and CEO of Soteria Biotech Co., Ltd, Mr. Hung Ta Hsiao, to discuss implementation strategies and answer any questions.

Soteria One Point Zero offers a unique opportunity to revolutionize healthcare delivery in Estonia by addressing critical challenges in OSA diagnosis and treatment. We are confident that this collaboration can significantly improve patient outcomes while optimizing healthcare resources.

Thank you for considering our proposal. We look forward to the possibility of working together to advance healthcare standards in Estonia.

Yours sincerely, Vilnis Lībietis Member of the Board

THE DOCUMENT HAS BEEN SIGNED WITH A SECURE ELECTRONIC SIGNATURE AND CONTAINS A TIME STAMP.

Page 1 of 3

Manufacturer’s Declaration

in relation to Regulation (EU) 2023/607 amending Regulations (EU) 2017/745 and (EU) 2017/746 as

regards the transitional provisions for certain medical devices and in vitro diagnostic medical devices, in

particular with respect to

 the validity of certificates issued under Council Directive 90/385/EEC on Active Implantable Medical

Devices (AIMDD) or Council Directive 93/42/EEC on Medical Devices (MDD) (Directive Certificates)

and/or1

 the compliance of the devices and us as their manufacturer with the conditions for the continued

placing on the market and putting into service

Manufacturer name Soteria Biotech Co., Ltd.

Manufacturer address and contact details

12F, No.365 Fude 3rd Rd., Xizhi Dist.,

New Taipei City 22151

Taiwan

Single Registration Number (SRN) (if available) TW-MF-000013120

Authorised Representative name (if applicable) MT Promedt Consulting GmbH

Authorised Representative address and contact details

Ernst-Heckel-Straße 7

66386 St. Ingbert,

Germany

Single Registration Number (SRN) (if available) DE-AR-000000085

Notified body name (if applicable) TÜ V SÜ D Product Service GmbH

□ See attached schedule

Notified body number (if applicable) CE0123

□ See attached schedule

Directive Certificate number(s) to which this confirmation is made (if applicable)

G1 102623 0002 Rev. 00

□ See attached schedule

Original expiry date as indicated on the Directive Certificate prior to the extension of the validity (if applicable)

26 May 2024

□ See attached schedule

End date of extended validity/transition period 31 December 2028

□ See attached schedule

1 The first condition is not applicable in case of devices for which the conformity assessment procedure pursuant to MDD did not

require the involvement of a notified body, for which the declaration of conformity was drawn up prior to 26 May 2021 and for which the conformity assessment procedure pursuant to this Regulation requires the involvement of a notified body.

Page 2 of 3

We, as the manufacturer declare under our sole responsibility:

 for the above listed Directive Certificate (or see attached schedule, if multiple certificates) the

conditions for the legal extension of validity as required in Article 120.2 of the MDR are met and/or2

 the listed device(s) in the attached schedule and we as their manufacturer are in compliance with

the conditions listed in Article 120.3c of the MDR for continued placing on the market and putting

into service,

namely by fulfilling the following conditions:

 Directive Certificate(s) as listed above or in the attached schedule

 Directive Certificate(s) covering the listed device(s) was/were issued after 25 May 2017, was/were

valid on 26 May 2021 and have not been withdrawn afterwards.

Expired/expires after 20 March 2023:

Formal application(s) to the notified body in accordance with Section 4.3, first subpara-

graph of Annex VII MDR for conformity assessment has/have been made or will be

made/submitted by us to a notified body no later than 26 May 2024 for the device(s) listed

in the attached schedule or its/their substitute(s) and signed written agreement(s) is/will be

in place in accordance with Section 4.3, second subparagraph of Annex VII MDR before

26 September 2024.

 Quality Management System (QMS)

A QMS in accordance with Article 10(9) MDR will be put in place by no later than 26 May 2024.

 Device(s) as listed in the attached schedule

 The device(s) continue to comply with the AIMDD or MDD.

 There are no significant changes in the design and intended purpose.

 The device(s) do not present an unacceptable risk to health or safety of patients, users or other

persons, or to other aspects of the protection of public health.

Signed for and on behalf of the manufacturer:

Soteria Biotech Co., Ltd.

New Taipei City, Taiwan January 4, 2024

Hung Ta Hsiao, CEO

[email protected]

2 The first condition is not applicable in case of devices for which the conformity assessment procedure pursuant to MDD did not

require the involvement of a notified body, for which the declaration of conformity was drawn up prior to 26 May 2021 and for which the conformity assessment procedure pursuant to this Regulation requires the involvement of a notified body

Page 3 of 3

Schedule of Devices The above Manufacturer’s Declaration is valid for the following devices:

Identification of the device(s)3 (e.g., device name, family/group name device model or catalogue number)

Directive Certificate number(s) to which this confirmation is made (if applicable)

Original expiry date as indicated on the Directive Certificate (s) prior to the extension of the validity (if applicable)

Notified Body name and number that issued the Directive Certificate (if applicable)

Notified Body name and number where the MDR application was lodged/contract signed (if applicable)

End date of extended validity / transition period

Substitute Device(s) (if applicable)

Soteria One Point Zero

G1 102623 0002 Rev. 00

26 May 2024 TÜ V SÜ D Product Service GmbH CE0123

31 December 2028 N/A

3 for devices with AIMDD/MDD certificate(s) the identification should be as in the certificate, and only if the certificate has a generic scope it should be as defined

above)

Soteria Biotech Co., Ltd.
 2024.07.22

CT / CBCT
 Imaging Requirements for all Customers

Suitable CBCT Examples *or any available CT is suitable

Viso® family of premium CBCT imaging units Viso® family of premium CBCT imaging units Panoramic X-ray system eco-x

Standard CT Image (Hospitals)

• Head and Neck CT


• At least the area in the red box


• From the circled nasal cavity and down to the hyoid bone

Pixel size: 0.0625cm x 0.0625cm x 0.0625cm or better

Standard CBCT Image (Dental) Pixel size: 0.0625cm x 0.0625cm x 0.0625cm or better Shooting range: FOV > 10cm x 10cm or greater

• Head and Neck CT


• At least the area in the red box


• From above the end of the circled hard palate and down to the hyoid bone

CT / CBCT Shooting Method

1. Before the shooting, the patient should: • lie down (CT) or sit/stand (CBCT), • be in centric occlusion, • completely exhale all air from chest

and abdomen.


2. During the shooting, the patient should: • keep holding exhaled state, • strictly avoid swallowing.

*contrast agent isn’t required

1

Revision History

Revision No. Revision Description Publication Date Author

01 Initial release 2019/05/02 Hsiao

2

Contents

1. Introduction ............................................................................................................................................. 3 2. Purpose of the report ............................................................................................................................ 3 3. Product description ................................................................................................................................ 3 4. Intended use ........................................................................................................................................... 3 5. Indication for use .................................................................................................................................... 3 6. Method ..................................................................................................................................................... 4 7. Selection and exclusion of Subjects ................................................................................................... 6 8. The specification of CBCT scanner .................................................................................................... 7 9. Assessment of Efficacy ......................................................................................................................... 7 10. Statistics .................................................................................................................................................. 8 11. Assessment of Saftey ........................................................................................................................... 8 12. Results ..................................................................................................................................................... 9 13. Conclusion ............................................................................................................................................ 11 14. Reference .............................................................................................................................................. 11

3

1. Introduction

This document describes the performance of Soteria One Point Zero (version no.: 1.0). This

report is prepared with the consideration of following references:

– Software as a Medical Device (SAMD): Clinical Evaluation – Guidance for Industry and

Food and Drug Administration Staff (December 8, 2017);

– Clinical Evaluation: A Guide for Manufacturers and Notified Bodies Under Directives

93/42/EEC and 90/385/ECC (MEDDEV 2.7/1 revision 4)

2. Purpose of the report

This performance report is a late-development-stage report. The goal is to establishing a

baseline of the product performance and as an evidence that Soteria One Point Zero can be

used safely and efficiently.

3. Product description

Soteria One Point Zero is a cloud based stand-alone software platform designed to display and

analyze head and neck medical images. The software utilizes automatic airway detection

algorithm / manual airway selection, 3-D model reconstruction, and computational fluid

dynamics (CFD) to simulate and calculate for flow characteristics include flow velocity and

pressure within the airway. These information can be used to determine if the airway is

obstructed and at which location.

4. Intended use

Soteria One Point Zero is a cloud based stand-alone software platform designed to display and

analyze head and neck medical images.

5. Indication for use

Soteria One Point Zero is a cloud based stand-alone software platform designed to display and

analyze head and neck medical images (DICOM compatible images) for airway flow

4

characteristics including flow velocity and pressure within the airway. These results are intended

to be used along with patient’s clinical history, as well as physicians’ professional judgment.

These information are not for immediate diagnostic nor treatment use, they are used to

determine if the airway is obstructed and at which location.

6. Method

This study included 200 subjects from different sites, sites include Chang-Gang Memorial

Hospital (Taoyuan, Taiwan, with 89 subjects), China Medical University Hospital (Taichung,

Taiwan, with 71 subjects), and Kaohsiung Medical University Hospital (Kaohsiung, Taiwan,

with 40 subjects). Subjects were grouped into control group (subject without obstructive

sleep apnea, OSA) and OSA group by the diagnosis.

Obstructive sleep apnea is a common sleep disorder of adults and children, it is

characterized by repetitive transient reversible upper airway obstructions during sleep.

In general, patient with OSA has narrower upper airway compare with subject without

OSA. According to American Academy of Sleep Medicine (AASM), OSA is diagnosed by

using the polysomnography (PSG) result plus clinical symptoms, this is currently the gold

standard for OSA diagnosis. The result of PSG is presented as apnea hypopnea index or

AHI and it is defined as number of apnea (complete cessation of airflow for at least 10

seconds) or hypopnea (airflow decreases by 50% for 10 seconds or decreases by 30% if

there is an associated decrease in the oxygen saturation or an arousal from sleep) in an

hour. The severity of OSA is classified by AHI where AHI < 5 per hour is considered as

normal or minimal, AHI ≥ 5, but < 15 per hour is mild, AHI ≥ 15, but < 30 per hour is

moderate, and AHI ≥ 30 per hour is considered as severe.

The study procedure flow chart is shown in Figure 1, subject first get diagnosis by the first

clinician with gold standard method. The same subject then takes the Cone Beam

Computed Tomography (CBCT) scan, the scanned image is analyzed by the software and

the report is presented to the second clinician and the clinician determines if subject’s

5

airway is obstructed. Finally the results from the first and the second clinicians are

compared.

The specification of CBCT scan is shown in Figure 2.

Figure 1. Study flow chart

Figure 2. The region of CBCT scan. The red rectangle region is around 7-9cm; Pixel size: 0.06cm * 0.06cm * 0.06cm;

6

The criteria for determine subject with or without airway obstruction is shown in Table 1 below:

Table 1. Criteria for determine subject with or without airway obstruction

Software report Condition

The color distribution shows a uniform color which means minimal pressure or velocity change. The interpretation is the air flow through this section of the airway smoothly without obstruction Condition = Without obstruction

The color distribution showed color difference at the section boxed in red which means pressure difference at that location. The interpretation is the air flow through that section with some obstruction causing change of the pressure and velocity. Condition = Airway obstruction

The color distribution showed a sharp color change from green to dark blue at the section boxed in red which means a large pressure difference at that location. The interpretation is the air flow through this section of the airway with obstruction causing sharp change of the pressure and velocity. Condition = Airway obstruction

7. Selection and Exclusion of Subjects

Selection Criteria: 1. Male or female subjects age equal to or above 20, equal to or below 65 2. Polysomnography (PSG) study, with study time for at least 5 hours or above 3. BMI equal to or less than 30

7

Exclusion Criteria: 1. Severe uncontrollable cardiovascular, neurological or metabolic diseases 2. Cannot take x-ray examination 3. Pregnant or possible pregnant women 4. Central sleep apnea 5. Subject had head and neck surgery (may affect the scanning result)

Discontinuation of study: If subject is not feeling well during any part of study, subject can decide to terminate the study

8. The specification of CBCT scanner

The specification of CBCT scanner used is shown in Table 2.

Table 2. Specification of CBCT scanner

X-ray Tube

Voltage 60~120kV

Current 1~20mA

Focal Spot Size 0.3~0.5mm Signal Gray Scale 12~16bit Image Detector <40cm×40cm Source to Detector Distance >50cm Source to Patient Distance >48cm Field Of View(FOV) (6~23)×(6~23) Scan Time(s) <30 Image Acquisition 180°~360° Reconstruction Time(s) <120 DICOM File Size 50M~2G Voxel Size <0.4mm×0.4mm×0.4mm

9. Assessment of Efficacy

This study is to use the criteria shown in Table 1 to determine if the airway of the subject is

obstructed. The comparison is done by compare the clinician determined result using the

8

software vs. gold standard diagnostic result. If the accuracy of the result determined by

clinician using the software compared with gold standard diagnostic result is 70% or above, it

is considered as effective.

10. Statistics

The validity of the index test is measured with sensitivity and specificity shown in two-by two

table below

Table 3. 2x2 table

Cell “a” we enter those who have positive results for the index test and the gold standard test

also tested positive. These are the true positive.

Cell “b” we enter those who have positive results for the index test but do not test positive

according to the gold standard test. These are false positive.

Cell “c” we enter those who have negative results for the index test but positive results on the

gold standard test. These are false negative.

Cell “d” we enter those who have negative results for the index test and the gold standard test

also test negative. These are the true negative.

Sensitivity is the ability of a test to correctly identify an individual as positive, it is also called

true positive rate. Sensitivity= a/(a+c) which is probability of being tested positive when disease

is present. Specificity is the ability of a test to correctly identify an individual as negative, it is

also called true negative rate. Specificity= d/(b+d) which is probability of being tested negative

when disease is absent. Accuracy is the sum of the true positive and the true negative among

all tested. Accuracy = a+d/(a+b+c+d).

9

11. Assessment of Safety

The specification of CBCT is shown in table 2. In accordance with the scanning area (Figure 2), the maximum dosage per scan is approximately 120μSv. According to the international commission on radiological protection, ICRP, the limitation of radiation dosage for an adult is 5 mSv (5000μSv ) per year. Therefore, as long as subject does not expose to radiation exceeding the mentioned limitation, there should be no harm from the radiation.

12. Results

Figure 3 shows the air flow pressure color distribution of subject’s airway with AHI of 7.9, 13.5,

and 81. As AHI gets higher, the pressure difference (color difference) is also higher. It can clearly

seen the largest pressure difference is at the location where the airway is narrowest (boxed in

red).

Figure 3. The airflow pressure color distribution of subjects with different AHI

Table 4 shows the relative physiological data, subjects are divided into Group 1 (AHI<5) 63

subjects, Group 2 (5≦AHI<15) 8 subjects, Group 3 (15≦AHI<30) 42 subjects, and Group 4

(AHI≧30) 87 subjects.

10

Table 4. The relative physiological data of subjects

Group1(AHI<5) Group2(5<AHI<15) Group3(15<AHI<30) Group4(AHI>30)

case 63 8 42 87

Sex Male 42 6 33 58

Female 21 2 9 29

Age 49.4±12.2 39.4 ±12.7 49.3 ±12.2 54.3 ±12.5

AHI 1.9 ±1.5

8.9 ±3.3 20.7±4.1 55.2 ± 15.7

Table 5 shows that doctor use the report from Soteria One Point Zero to evaluate whether

subject’s airway has obstruction and got sensitivity, specificity, and accuracy above 80%.

Table 5. Compare PSG (gold standard) with doctor use the report from Soteria One Point Zero

PSG Diagnosis

(AHI≧5 is positive)

Positive

(AHI≧5)

Negative

(AHI<5) Row Total

Doctor’s evaluation use report from

Soteria One Point Zero

Positive 121 8 129

Negative 16 55 71

Column Total 137 63 200

Sensitivity=121/137=0.88

Specificity=55/63=0.87

Accuracy =（121+55）/200=0.88

11

13. Conclusion

This study is a blind test with OSA patient and we conclude the report from Soteria One Point

Zero is helpful to doctors to determine if subjects has airway obstruction and with accuracy of

88%.

14. Reference

Caples, S. M., T. Kara, et al. (2005). "Cardiopulmonary consequences of obstructive sleep apnea." Semin

Respir Crit Care Med 26(1): 25-32.

El-Ad, B. and P. Lavie (2005). "Effect of sleep apnea on cognition and mood." Int Rev Psychiatry 17(4):

277-82.

Fogel, R. B., A. Malhotra, et al. (2004). "Sleep. 2: pathophysiology of obstructive sleep apnoea/hypopnoea

syndrome." Thorax 59(2): 159-63.

Katz, I. M., B. M. Davis, et al. (1999). "A numerical study of particle motion within the human larynx and

trachea." J Aerosol Sci 30: 11.

Keyhani, K., P. W. Scherer, et al. (1995). "Numerical simulation of airflow in the human nasal cavity." J

Biomech Eng 117(4): 429-41.

Lam, B., C. G. Ooi, et al. (2004). "Computed tomographic evaluation of the role of craniofacial and upper

airway morphology in obstructive sleep apnea in Chinese." Respir Med 98(4): 301-7.

Leiter, J. C. (1996). "Upper airway shape: Is it important in the pathogenesis of obstructive sleep apnea?" 、

Am J Respir Crit Care Med 153(3): 894-8.

Li, W. I., M. Perzl, et al. (1996). "Aerodynamics and aerosol particle deaggregation phenomena in model

oral-pharyngeal cavities." J Aerosol Sci 27: 18.

Liao, Y. F., C. S. Huang, et al. (2003). "The utility of cephalometry with the Muller maneuver in evaluating

the upper airway and its surrounding structures in Chinese patients with sleep-disordered breathing."

Laryngoscope 113(4): 614-9.

Malhotra, A., Y. Huang, et al. (2002). "The male predisposition to pharyngeal collapse: importance of airway

length." Am J Respir Crit Care Med 166(10): 1388-95.

Pae, E. K., A. A. Lowe, et al. (1997). "A role of pharyngeal length in obstructive sleep apnea patients." Am J

Orthod Dentofacial Orthop 111(1): 12-7.

Polo, O. J., M. Tafti, et al. (1991). "Why don't all heavy snorers have obstructive sleep apnea?" Am Rev

Respir Dis 143(6): 1288-93.

Remmers, J. E., W. J. deGroot, et al. (1978). "Pathogenesis of upper airway occlusion during sleep." J Appl

Physiol 44(6): 931-8.

Schwab, R. J. (1996). "Properties of tissues surrounding the upper airway." Sleep 19(10 Suppl): S170-4.

Schwab, R. J., W. B. Gefter, et al. (1993). "Dynamic upper airway imaging during awake respiration in

normal subjects and patients with sleep disordered breathing." Am Rev Respir Dis 148(5): 1385-400.

Schwab, R. J., W. B. Gefter, et al. (1993). "Dynamic imaging of the upper airway during respiration in

normal subjects." J Appl Physiol 74(4): 1504-14.

Stierer, T. and N. M. Punjabi (2005). "Demographics and diagnosis of obstructive sleep apnea." Anesthesiol

12

Clin North America 23(3): 405-20, v.

Subramaniam, R. P., R. B. Richardson, et al. (1998). "Computational fluid dynamics simulations of

inspiratory airflow in the human nose and nasopharynx." Inhalation Toxicology 10(5): 30.

Vos, W., J. De Backer, et al. (2007). "Correlation between severity of sleep apnea and upper airway

morphology based on advanced anatomical and functional imaging." J Biomech 40(10): 2207-13.

Useful Q&A on Soteria AI platform:

Q: How many OSA cases are being operated by Soteria platform for AI system training at the moment [September 2022]? Is there any post-operative feedback optimization training?

A: Soteria AI system training is currently based on a total of 6872 cases (Taiwan 6840, the UK 20, the Netherlands 12). When the amount of cases had reached 200, the OSA Report accuracy rate reached 88%. Then Soteria successfully passed the TFDA and CE certification. The missing percentage is mainly mild OSA cases, since it’s challenging to collect data from patients with mild symptoms. The accuracy rate of post-treatment optimization training is currently 90% (published at the 2018 Asian Sleep Conference).

Q: How many patients have been successfully assisted by Soteria AI platform at the moment? How did it become possible to predict post-treatment improvements with such high efficacy?

A: Soteria has registered 6840 patients’ upper airways improvement, based on provided clinical feedback in Taiwan. Initially, for the purpose of TFDA and CE certification, Soteria has participated in 200+ PSG diagnostics in sleep centers. PSG complete output data was fed to the AI program together with pre-treatment CT images of the same patients to detect corresponding information. Post-treatment CT images of the same patients followed up feeding the AI program to further ‘teach’ it to recognize results and predict them with just the input of CT images only. Since then, by fulfilling the contract, Kaohsiung Medical University Hospital (Kaohsiung city, Taiwan) is continuing to provide pre- and post-treatment detailed clinical feedback to Soteria to refine and advance our medical software. Also, this key corresponding information uncovered by the system can’t be seen by a doctor on a CT scan, it can only be simulated and calculated by the AI and offered as visual colorful image of airflow pressure and velocity. Basically, now a doctor only needs to input CT image, and receive complete+ gold standard diagnostic result without asking a patient to stay in a sleep center.

Q: Will the treatment methods/ standard definitions/ rehabilitation definitions be different in different countries and regions? Will the above definitions or methods have a certain influence on the judgment capability of Soteria AI platform?

A: For OSA, these treatment methods/ standard definitions/ rehabilitation definitions in different countries are all guided by the criteria of the American Sleep Society and the European Sleep Society, so there is no difference.

At present, there are no differences detected among different human races in the OSA test results conducted in the Netherlands and the UK. Due to the uniqueness of Soteria AI+CFD technology, the analysis result is based on breathing pressure and breathing speed, not on the conventional respiratory tract volume and cross-sectional area. Thus, belonging to a different human race creates zero influence on Soteria OSA evaluation capability. In other words, the software can be efficiently applicable for any patient.

Q: In addition to the FDA and CE certification, does the AI platform have any exclusive doctor/university certification?

A: Currently, thoracologists, otolaryngologists, and neurologists in Taiwan are all Soteria platform users and have published medical journal papers for certification.

Q: Are there any requirements for CT scanning while standing/lying down, or leaning forward/backward/sideways, etc.?

A: When scanning while lying down, patient should lie flat horizontally. When scanning while standing, patient should stand up straight. Patient shouldn’t move/tilt his/her head. Patient should completely exhale, then radiologist starts scanning.

Q: How many times CT scans are usually performed for different recommended solutions? If the same case requires multiple scans and analyses, is the cost being also charged for a single case?

A: There are mainly these 3 different treatment plans: 1.Respirator (CPAP): only needs to scan once —> single charge. 2.Oral Sleep Device: two shots before treatment —> one case —> single charge. 3.Orthodontic/Laryngeal Surgery (MMA): once before the operation, once after the operation —> the cost of the case is charged twice.

Q: What level of details can AI platform possibly provide with the guidelines/risk warning? For example, will it only be “Surgery is required”, or something like "Tracheal dilation surgery is required (the cut part is marked in the red area of the image), pay attention to the adhesion of the lower part" and the like?

A: Soteria report mainly provides the location of the respiratory tract obstruction, the severity of the obstruction, and the cause of the blockage. The detailed guidelines (as in the example above) will not be recommended by AI platform. The physician proposes the final solution with the help of all information provided in the report.

Q: Is it convenient to provide a test account to understand the entire platform operation process?

A: Sure. Just click https://lydata.com.tw/lydata_soteriabio/admin/index.html and give us a call to get your login and password.

Q: Is it convenient to provide instructions for making standard CT images for reference?

A: Sure. Please refer to the corresponding file.

Q: Will there be some public information, brochures or posters that can be provided to the hospitals to exhibit/post/introduce to patients?

A: Yes, Soteria can provide you with digital copies for printing.

Q: If patient needs to use oral sleep device, does Soteria also provide them, or can the local hospitals make these braces themselves?

A: Soteria can also provide oral sleep devices, which in turn require intraoral scan files and CT files, and additional business negotiations and a whole new business setup. Or the local hospitals can make their own braces.

Q: If patient has undergone surgery and the trachea is in unnatural state, does the postoperative situation apply to this AI evaluation?

A: Yes, it is applicable. AI platform can also provide an analysis report on the degree of narrowing of the respiratory tract, and the rest of the information in the same way as for a natural state.

Q: Will the AI platform store records of patient data? If so, where will they be stored (server location)?

A: No, Soteria doesn’t store records of patient data. In accordance with medical regulations and ISO13485 regulations, the uploaded image files will first be stored in a cloud company that meets the information security specifications. European data is stored in Amsterdam (Netherlands), Taiwanese data is stored in Hsinchu (Taiwan).

Patient OSA analysis data (Report) will be automatically deleted in two days after it has been downloaded by the hospital.

Q: How long are the Reports stored on the platform?

A: In accordance with medical regulations and ISO13485 regulations, patient OSA analysis data (Report) will be automatically deleted in two days after it has been downloaded by the hospital. As long as there's business with the hospital, hospital can choose to store used reports on the platform. However, based on Soteria experience, hospitals rather prefer to keep downloaded reports (PDF format) in the hospital. If business with the hospital is terminated, hospital will still be granted with some time to save all paid reports as PDF files, and then all relevant data will be deleted.

Some straight-forward FAQs from doctors’ perspective:


 Q1: Sounds like you are trying to load me with something new, that requires to get studied, get used to... And I’m already way beyond busy!

A1: On the contrary, Soteria is designed to UNLOAD your hectic schedule. Expert’s time and direct attention to patients are most precious. The rest of the routine that required 1-3 days per visitor, now can be done in 1-2 hours. Just drag and drop CT scans to our online platform, and receive OSA analysis Report.
 What Doctor DOES see on Soteria platform: it processes uploaded CT images and issues a Report with the same but colored CT images and detailed crucial data, exceeding PSG’s.
 What Doctor DOES NOT see on Soteria platform (it’s background functioning): our system is “trained” to simulate the work of upper respiratory tract during sleep WITHOUT the need for the Patient to actually sleep in the laboratory.
 The complete process is elegantly simplified for your ultimate convenience!
 In other words, WHY should more and more Patients continue sleeping in laboratories, when thousands and thousands of other Patients with different degrees of OSA complexity have already slept before them?
 WHY spending 1-3 days on one Patient, when for the same retail value and with great accuracy you can examine him/her in an hour or two? And our Report will show the result of simulation, how Patient's upper airways behaved while “sleeping”.

Q2: Are you trying to replace Doctors with AI? :)

A2: No, it’s illegal, and/or must be so. Doctor’s expertise is INDISPENSABLE! The software does NOT make decisions INSTEAD of a doctor, it only analyzes big data by just being a "smart calculator”.


The software removes 80% of work routine: collecting PSG/other measurements, data entry, simulation, calculation, analysis; and outputs visually intuitive (also patient- friendly) data.
 20% of the work can only be done by an experienced doctor, relying on the Report to choose the best method of treatment. And these 20% are most important.

Q3: We’ve (just) invested a ton in the best sleep center equipment! And now you want us to replace Gold Standard PSG with AI?

A3: At that moment it surely was your best investment in business growth and service quality. Today, why not to have both at your disposal? You already have a certain number of beds in your sleep laboratory. How about “installing” more “beds”, even more “sleep centers” with a small fraction of their cost?
 These virtual “beds” DON’T require: space in your clinic, more equipment and consumables, maintenance, new training, qualified personnel, extra services for Patients during the whole procedure, extra attention from Doctors, etc.
 INSTEAD, these virtual “beds” provide you with precise certified analyses AND attract largely underexamined Patients who did not want to come previously, because:
 --clinics don’t offer quick diagnosis,
 --it takes long waiting time to finally get examined in the laboratory,
 --they’re busy and can't set aside as much as 1-3 days for this procedure,
 --their OSA has unnoticeable or light form, which they believe is unworthy of 1-3 days spent in a hospital,
 --they feel it’s too complex and morally uncomfortable,
 --they hesitate to visit any medical institution due to fear of Covid-19 or other contagious diseases,
 --they are children (including: they’re just too small; with lagging development; with hysterical behavior; underachieving pupils with ADHD; who require orthodontic treatment), 
 --they are old people, disabled (with caretakers), weak (Patients who need to identify whether OSA is one of the root reasons why they’re sick for a long time, don’t recover well, recover poorly after surgery).
 All these categories of Patients are not even included in the clinical OSA studies, and thus this market is largely underpenetrated.
 So, now you see how you can rethink your service quality and business EFFICIENCY with Soteria’s improved approach. Our software helps to EXPAND your business without inserting a stick into the wheel of what has already been established by your esteemed clinic.

Q4: Why do we need your AI system if Pulmonologist can see where the narrowing is on the CT scan, and whether surgery is needed or not?

A4: Of course, complex and trouble-free cases are always obvious for Pulmonologist, but all in between – not so much.
 Our system is especially good when it’s difficult for а Doctor to determine by eye whether there is an OSA. That’s why Soteria has created visual Reports: air flows are

highlighted in colors, how they move, where they get stuck, pressure, velocity, etc. – all during Patient’s simulated "sleep”.
 Doctor can’t possibly SEE how Patient’s upper airways behave when he snores and suffocates. AI can. This process can only be “SEEN” with Computational Fluid Dynamics (CFD) simulation; with PSG/else – it’s just a bunch of numbers (with some inconsistency).
 In 2016–2017, the medical software Developer, together with leading Pulmonologists in Taiwan, conducted over 200 comprehensive measurements and studies in the sleep centers of Taiwan's leading hospitals to collect high quality big data from a wide variety of OSA cases, and "teach" AI to "see" all changes in the Patients’ breathing patterns in their sleep.
 Nowadays, the world's first patented and CE certified system is operating thousands of such cases, constantly self-learning and enhancing its Reports’ precision beyond current 88%–91%.

Q5: Can you provide contacts of Doctors who have already implemented your system? We’d like to personally hear their feedback.

A5: Currently, there’re over 15 hospitals, medical centers and clinics in Taiwan, Singapore, the Netherlands, and the UK, that are successfully using Soteria platform and satisfied with OSA Reports.
 Please send us a message to [email protected] or give us a call at… (WhatsApp, Telegram), and we’ll arrange a conversation with an available English-speaking doctor without breaching existing NDAs (Non-Disclosure Agreements).