(rev 2023.06.07)
| ENGLISH |
Model 1000
Wellness Oximeter
Device Description
The Algo SaMD™ oximeter is a wellness oximeter configured to operate in Software as a Medical Device (“SaMD”). The software device is designed to run remotely as an optional service component of a complete all.health ICM™ system, a non-medical-device clinical decision support system. The software collects input from compatible remote hardware bands or other sensors, then filters, analyzes, and validates the collected data to produce a clinical-grade pulse measurement of heart rate (HRp), respiration rate (RRp), and blood oxygen (SpO2%) for output, including outlier rejection, averages, trends, and ranges over time for the purpose of remote patient monitoring.
Intended Use |
The Algo SaMD™oximeter is intended for the noninvasive spot-checking or continuous remote monitoring of trends in well-perfused adult patients in a non-clinical environment, outside of the hospital, clinic, or transport setting.The Algo SaMD™ oximeter is indicated for collecting remote data, then filtering, validating, and analyzing the collected data to output a clinical measurement, including outlier rejection, averages, trends, and ranges over time, of:
-- Heart Rate (HRp) during both no-motion and motion conditions,
-- Respiration Rate (RRp) during no-motion conditions, and,
-- Blood Oxygen (SpO2%), a measure of functional oxygenation saturation of arterial hemoglobin. blood, during no-motion conditions.
The Algo SaMD™ oximeter is a Class II (enforcement discretion) medical device.
The Algo SaMD™ oximeter is a Class II (enforcement discretion) medical device.
Federal law (USA) restricts this device to sale by or on the order of a licensed practitioner.
Contraindications |
• There are no known contraindications for the use of Algo SaMD™ oximeter software.
Warnings
Cautions
• There are no known contraindications for the use of Algo SaMD™ oximeter software.
Warnings
• This device is NOT INTENDED FOR EMERGENCY RESPONSE, critical care, or urgent care use. • This device is not intended for real-time use. Spot-check results can be delayed by five minutes, while trended measurements and notifications have trending and processing steps, and can be delayed by hours to days.
• This device for remote use. Data must be transmitted prior to collection and analysis.
• There are no alarms generated by this device.
• There are no alerts generated by this device.
• It is the patient’s responsibility to contact a qualified healthcare provider or call 911 (or the appropriate emergency response number) for health emergencies or urgent concerns.
• This device for remote use. Data must be transmitted prior to collection and analysis.
• There are no alarms generated by this device.
• There are no alerts generated by this device.
• It is the patient’s responsibility to contact a qualified healthcare provider or call 911 (or the appropriate emergency response number) for health emergencies or urgent concerns.
Cautions
• Providers should verify unexpected readings with another device.
• Clinical correlation between a patient’s condition and displayed results is recommended.
• Factors that can degrade any pulse oximeter’s accuracy include:
• Clinical correlation between a patient’s condition and displayed results is recommended.
• Factors that can degrade any pulse oximeter’s accuracy include:
-- excessive ambient light on the sensor or body,
-- excessive body motion,
-- very low blood pressure, poor pulses, cold air or hands, restrictive cuffs or clothes, -- improperly worn or excessively loose or tight sensor,
-- carboxyhemoglobin, methemoglobin, or dyes in the blood,
-- paint, ink, tattoos, or dyes on the skin in the light path of the sensor,
-- scars, deep skin pigmentation, or excessive dirt in the light path of the sensor,
-- anemia, very low or high hemoglobin levels,
-- high altitude or air travel,
-- non-compatible hardware,
-- low or dead hardware battery,
-- forgetting to wear hardware sensor, and,
-- other known, published factors.
Notes |
• This device is an optional component of a complete all.health ICM™ system. This device is not sold separately.
• This device collects data from compatible sensors, including the optional Band 01™ sensor.
• This device collects sensor data sent to the cloud from the all.health app.
• Patients should wear or use any optional sensors or sensor bands as directed by their provider, including regular charging, cleaning, and updates.
• Patients can view their own results on the all.health app.
• Providers can view a patient’s results using the all.health portal
Preparing to Use the Device |
-- excessive body motion,
-- very low blood pressure, poor pulses, cold air or hands, restrictive cuffs or clothes, -- improperly worn or excessively loose or tight sensor,
-- carboxyhemoglobin, methemoglobin, or dyes in the blood,
-- paint, ink, tattoos, or dyes on the skin in the light path of the sensor,
-- scars, deep skin pigmentation, or excessive dirt in the light path of the sensor,
-- anemia, very low or high hemoglobin levels,
-- high altitude or air travel,
-- non-compatible hardware,
-- low or dead hardware battery,
-- forgetting to wear hardware sensor, and,
-- other known, published factors.
Notes |
• This device is an optional component of a complete all.health ICM™ system. This device is not sold separately.
• This device collects data from compatible sensors, including the optional Band 01™ sensor.
• This device collects sensor data sent to the cloud from the all.health app.
• Patients should wear or use any optional sensors or sensor bands as directed by their provider, including regular charging, cleaning, and updates.
• Patients can view their own results on the all.health app.
• Providers can view a patient’s results using the all.health portal
Preparing to Use the Device |
1 |
Download the Latest App
2 |
Select a compatible sensor.
The Algo SaMD™ software device is designed to be compatible with the all.health wrist-wearable Band 01™ sensor band (or equivalent).
Alternatively, a patient may use any recognized, compatible medical pulse oximeter that stores data in Apple Health or Google Fit on the patient’s smartphone, provided that the patient also grants permission for their data to be transmitted and collected by the all.health app (see Table 1, below).
Table 1. Sensors compatible with the Algo SaMD™ medical device
Summary of Human Subjects Studies |
A. Blood Oxygen Studies
IRB Human Subjects Protocols
Publications
The initial papers described the basis of multispectral oximetry (multi-color, multi-wavelength).
Multispectral approaches are known to be able to correct certain biases known to arise in pulse oximetry. Such biases were raised in the FDA communication updated November 7, 2022, and entitled “Pulse Oximeter Accuracy and Limitations: FDA Safety Communication.”
In the following human studies data, the multispectral approach is shown to be able to detect and automatically correct skin pigment and thickness bias in oximetry when sufficient wavelengths are available. Despite different thicknesses and spectral appearance of 5 tissues types (shown in graph A), these are corrected to leave only true blood spectra with multispectral differential spectroscopy (shown in graph B):
Download the Latest App
The app should be installed on the patient’s smartphone or tablet (if not, see below).
Compatible sensors should be paired to the patient’s smartphone (if not, see pairing in the app). If a 3rd-party sensor is to be used (such as a scale or glucose meter), read and follow the manufacturer’s instructions for connecting their sensor to Apple Health or to Google Fit.
Compatible sensors should be paired to the patient’s smartphone (if not, see pairing in the app). If a 3rd-party sensor is to be used (such as a scale or glucose meter), read and follow the manufacturer’s instructions for connecting their sensor to Apple Health or to Google Fit.
|
Download all.health App For iPhone
|
|
Download all.health App For Android
|
Select a compatible sensor.
The Algo SaMD™ software device is designed to be compatible with the all.health wrist-wearable Band 01™ sensor band (or equivalent).
Alternatively, a patient may use any recognized, compatible medical pulse oximeter that stores data in Apple Health or Google Fit on the patient’s smartphone, provided that the patient also grants permission for their data to be transmitted and collected by the all.health app (see Table 1, below).
Table 1. Sensors compatible with the Algo SaMD™ medical device
|
Brand / Model
|
Clinical Grade?
|
Data Collection Frequency
|
Data Collection Method
|
|
all.health Compatible Sensors
|
|||
|
all.health / Band 01™
|
Yes ***
|
Continuous***** Day and Night, Spot Check
|
via
all.health app |
|
Third-party Compatible Sensors
|
|||
|
Masimo / MightySat Rx Fingertip Pulse Oximeter*
|
Yes ***
|
Continuous***** and Episodic Use, Spot Check
|
via
Apple Health or Google Fit, and all.health app |
|
Apple / Watch **
|
No ****
|
Continuous***** Day and Night, Spot Check
|
via
Apple Health and all.health app |
Footnotes for Table 1:
* Or equivalent clinical-grade pulse oximeter with heart and respiration rate.
** Or equivalent non-clinical-grade pulse oximeter with heart and respiration rate.
*** Data filtered and validated is marked as clinical-grade by Algo SaMDTM.
**** Non-clinical hardware devices produce data that cannot be labeled clinical-grade by Algo SaMD™.
This may restrict the use of such data by healthcare providers in clinical care.
Non-clinical-grade data is collected by other all.health software services and displayed as consumer-grade data.
***** Continuous in this context means that the sensor data collected is continuous, such as the Band 01™ sensor band will collect data every few seconds to every 5 minutes. However, because of the all.health ICM™ system design, this sensor data may not upload to the system cloud for hours or days, and may require further processing and analysis. Therefore the displayed results, based on continuous data, are not intended to be real-time.
3 |
Data collection is automatic.
Transfer of heart rate, respiration rate, and blood oxygen from compatible sensors to the app is automatic once the sensor is paired to the patient app, both whenever the patient app is opened or after 10-15 minutes have passed, provided the app is within 6 feet (2 meters) of the paired sensor or band sensor.
Uploading of heart rate, respiration rate, and blood oxygen from compatible sensors from the app to the ICM™ system cloud for collection, filtering, analysis, and validation by the Algo SaMD™ device software is automatic, provided a wi-fi, cellular, or all.health hub internet is available.
Data is validated and trended over time.
4 |
Data Viewing.
Data can be viewed as follows:
- ● Patients can view their heart rate, respiration rate, and blood oxygen in the patient app on the “data.health” view. Trends are displayed by tapping on any measurement tile.
● Providers can view a patient’s heart rate, respiration rate, and blood oxygen on the all.health provider portal. Trends can be viewed.
- FDA-regulated clinical-grade measurements are labeled “study verified” (displayed with 1 checkmark, “✓“) or “clinical-grade” (displayed with 2 checkmarks, “✓✓”).
- FDA-unregulated signals (steps, for example) are labeled “study verified” (displayed with 1 checkmark, “✓“) if a controlled, IRB-approved human study has verified the accuracy of the measurement.
- Unverified or other measurements are not displayed with trust checkmarks.
Performance Specifications |
Algo SaMD™ device software has been clinically validated in human studies under Institutional Review Board approval and Informed Consent to the following performance specifications:
Table 2: Performance Specifications
Algo SaMD™ device software has been clinically validated in human studies under Institutional Review Board approval and Informed Consent to the following performance specifications:
Table 2: Performance Specifications
| Feature | Performance Specifications |
| 1. Performance | |
| Heart Rate (HRp) | Range 20-220 No-motion: +/- 1 beat/min Motion: +/- 5 beat/min |
| Respiration Rate (RRp) | Range 5-30 No-motion: +/- 1 breath/min Motion: Not indicated |
| Blood Oxygen (SpO2%) | Range 80-100% No-motion, SpO2 94% and above: +/- 3% MAPE* No-motion, SpO2 80% to 93%: +/- 3.5% MAPE* No-motion, SpO2 79% and below: Not indicated Motion: Not indicated *MAPE = Mean Average Percentage Error |
| Modes of Operation | Continuous Data Collection Mode is a variable-duty mode for routine collection of data in 30-sec sample window, in 2 modes: ● Sample windows are collected every 5 minutes (“Normal” mode, 10% duty). ● Sample windows are collected every 30 seconds (“Raw Data” mode, 100% duty). Raw Data mode is used for certain analytics when higher fidelity data is needed for accuracy. Spot-check Data Collection Mode is an on-demand mode during no-motion conditions [to be enabled] |
| 2. Analysis and Validation | |
| Signal Validation | Validation Score (0-1). Score Ranges: High (>0.8), Med (0.6-0.8), Low (<0.6) confidence. |
| Signal Filtering | Reject low confidence signals. The validation score is used in the process of filtering and analyzing signals to mark them as clinical-grade. Signals of low confidence as well as outliers are rejected and filtered. |
| Signal Analysis | Average (HRp, RRp) or Median (SpO2%) values calculated over periods of sleep interval over 1 day, 1 week, 1 month Range (max, min) over periods of 1 hour and sleep interval. Trend values calculated over periods of 1 hour, 1 day, 1 week, 1 month |
| Risk Mitigation | Reject low confidence level signals Flag medium confidence level signals Validate only non-rejected signals |
| 3. Temperature | n/a (software) |
| 4. Humidity | n/a (software) |
| 5. Weight | n/a (software) |
| 6. Size | n/a (software) |
| 7. Electrical | n/a (software) |
Summary of Human Subjects Studies |
A. Blood Oxygen Studies
IRB Human Subjects Protocols
● POX-000
● POX-001
● POX-002
● POX-003
● POX-001
● POX-002
● POX-003
The initial papers described the basis of multispectral oximetry (multi-color, multi-wavelength).
Our multispectral approach was developed at Stanford University, and published in clinical journals for its accuracy in measuring blood oxygen, article here:
- Article Citation: Benaron, D. A., Parachikov, I. H., Cheong, W. F., Friedland, S., Rubinsky, B. E., Otten, D. M., ... & Kermit, E. L. (2005). Design of a visible-light spectroscopy clinical tissue oximeter. Journal of biomedical optics, 10(4), 044005-044005.
- Article Link:
https://www.spiedigitallibrary.org/journalArticle/Download?fullDOI=10.1117/1.1979504
In the following human studies data, the multispectral approach is shown to be able to detect and automatically correct skin pigment and thickness bias in oximetry when sufficient wavelengths are available. Despite different thicknesses and spectral appearance of 5 tissues types (shown in graph A), these are corrected to leave only true blood spectra with multispectral differential spectroscopy (shown in graph B):
From early peer-reviewed and published data, note in the Figure above that that hemoglobin is highly distorted and hidden in tissue scattering and baseline pigmentation in (A), while the use of multiple wavelengths allows extraction of accurate, accentuated hemoglobin spectra from this same set of tissues despite variation in skin, colon, and lip as shown in (B).
Clinical oximetry validation using multispectral analysis was first shown here:
- Article Citation: Continuous, noninvasive, and localized microvascular tissue oximetry using visible light spectroscopy. DA Benaron, IH Parachikov, S Friedland, R Soetikno, J Brock-Utne, et al. The Journal of the American Society of Anesthesiologists 100 (6), 1469-1475.
- Article link:
https://pubs.asahq.org/anesthesiology/article/100/6/1469/6613/Continuous-Noninvasive-and-Localized-Microvascular
Subsequent papers further validated the oxygenation results in vivo.
Initial Model Building
Initial Model Building
For clinical testing in the wrist-wearable Band 01™ sensor form. The system was tested in human volunteers using 4 LEDs. Testing involved model building using inspectable white-box algorithms. In a study of 34 people breathing a mixture of nitrogen and oxygen, saturation values varied between 78% to 100%. This produced approximately 1 hour of data or 3600 beats per subject for a total of about 122,400 samples. Each sample was compared to a ground truth of a clinical-grade Masimo oximeter. Model building was validated by N/2 build and N/2 analysis.
Each model was versioned, deployed, and validated in small samples (N=3-5) of performance in the complete system.
Submission Results:
Initial Model Building
[END OF IFU DOCUMENT]
An example of varying confidence is shown below (red line, 0-100 representing 0.00 to 1.00) vs reported oxygenation Spo2% (purple dot, 0-100). A wide variation in confidence can be seen at night in this single-patient overnight study, usually in response to motion artifact (excess motion is a rejection criterion in the Algo SaMD TM oximeter analysis).The creation of the confidence signal allows the automated validation and filtering, to obtain a more stable value.
Submission Results:
The following results were obtained, and provided as a basis for oximeter listing and equivalence:
Table 3: Blood Oxygen by Pulse
|
N Subjects
|
M samples
|
MAE%
|
Notes
|
|
18
|
1296 (est)
|
2.66%
|
Blood Oxygen
Basis for submission |
|
24
|
1728
|
1.25%
|
Improved oxygenation was demonstrated with improved rejection criteria,
|
More recent analysis shows some outliers improved over time by Median analysis. As we are seeking trends over hours to days, this filter is appropriate for trended data, but will not be used during sleep for low saturation detection during apnea (for which no averaging will occur between windows).
B. Heart Rate Studies
B. Heart Rate Studies
Human Studies Protocol
Publications
None
● SLE-000
● SLE-001
● ASU-001
● HRV-000
● HRV-001Publications
None
Initial Model Building
Heart rate by pressure-wave pulse detection (HRp) accuracy was tested using Band 01™ as follows:
- ● 64 recordings on 64 exercising test subjects x 1-2 hours each at all.health
- ● 20 recordings on 20 sleeping subjects x 8 hours each at all.health
- ● 55 exercising subjects at Arizona State University, during various physical activities including elliptical, running walking, cycling, and weight training
- ● 50 nocturnal recordings from the Stanford Sleep Clinic
for a total of 189 subjects in 594 recordings of subjects
As summary of the methodology, with visual samples to show the tools used, is as follows:
Submission Results
Submission Results
The following results were obtained, and provided as a basis for oximeter listing and equivalence:
The heart rate measurement model was coded and made to run in real-time to export tics (peak cardiac cycle time markers). Results were then exported for analysis. Post-submission monitoring continues before an updated Algorithm is released.
Table 4: Heart Rate by Pulse
Table 4: Heart Rate by Pulse
|
N Subjects
|
M samples
|
MAE%
|
Notes
|
|
189
|
594
|
2.2%
|
Basis for submission
|
C. Respiration Rate Studies
Human Studies Protocols
● RES-000 Devol Respiration Rate Study Protocol
● RES-001
● RES-002
Initial Model Building
The trials initially tested a combination of signals on a 1-wavelength optical wristband (Deion band, all.health Band 00™) to measure respiratory rate, then the models were revalidated in the 4-wavelength Band 01™ design.
Human Studies Protocols
● RES-000 Devol Respiration Rate Study Protocol
● RES-001
● RES-002
Initial Model Building
The trials initially tested a combination of signals on a 1-wavelength optical wristband (Deion band, all.health Band 00™) to measure respiratory rate, then the models were revalidated in the 4-wavelength Band 01™ design.
The goal of the trial was to demonstrate accuracy with a mean error of less than 1 breath/minute at rest, and less than 5 breaths/minute awake.
In this trial, we subjected healthy subjects to monitored respiration using clinical-grade chest straps. During a protocol that changed respiration over a range of values, subjects were monitored using wristbands as well as chest monitors.
The respiratory measurement model was coded and made to run in real-time to export tics (peak respiratory cycle time markers). Results were then exported for analysis.
Submission Results
In this trial, we subjected healthy subjects to monitored respiration using clinical-grade chest straps. During a protocol that changed respiration over a range of values, subjects were monitored using wristbands as well as chest monitors.
The respiratory measurement model was coded and made to run in real-time to export tics (peak respiratory cycle time markers). Results were then exported for analysis.
Submission Results
Table 5: Respiration Rate by Pulse
|
MAE
|
Notes
|
|
0.57 breaths/min
|
Respiration Rate by Pulse Basis for submission
|
Terms and Symbols |
The following terms and symbols are applicable to this device.
Table 6: Index of Terms
The following terms and symbols are applicable to this device.
Table 6: Index of Terms
|
Term
|
Definition
|
|
Band 00™
|
The all.health wrist-wearable research sensor band. For investigational use only.
|
|
Band 01™
|
The all.health wrist-wearable multispectral (4-wavelength, expandable) and acceleration sensor band platform.
|
|
Band 02™
|
The next generation all.health wrist-wearable (7-wavelength, expandable) and acceleration + gyro sensor band platform. For investigational use only.
|
|
Signal
|
A raw, time-varying signal from a sensor.
|
|
Ticks
|
A processed signal from a sensor, sent a processed rate, such as a heartbeat marker, or an averaged HR value every 5 minutes.
|
Table 7: Index of Symbols
Patients |
This IFU is intended for healthcare providers
• Please see the IFU for Patients in the Setting/About section of the patient app. • See FAQ / Troubleshooting in the Settings/About section of the patient app.
• Visit us on the web at www.all.health.
This IFU is intended for healthcare providers
• Please see the IFU for Patients in the Setting/About section of the patient app. • See FAQ / Troubleshooting in the Settings/About section of the patient app.
• Visit us on the web at www.all.health.
Troubleshooting, More Info, and Contact |
For More Information
• See IFU in the Setting/About section of the patient app or provider portal.
• See FAQ / Troubleshooting in the Settings/About section of the patient app or provider portal.
• Visit us on the web at www.all.health.
For Product Support or To Contact Us
• Connect directly to a customer support agent in the patient app or provider portal.
• Email us at info@all.health.
• Visit us on the web at www.all.health.
• Call us at 1.415.689.3018
Medical Device Info |
For More Information
• See IFU in the Setting/About section of the patient app or provider portal.
• See FAQ / Troubleshooting in the Settings/About section of the patient app or provider portal.
• Visit us on the web at www.all.health.
For Product Support or To Contact Us
• Connect directly to a customer support agent in the patient app or provider portal.
• Email us at info@all.health.
• Visit us on the web at www.all.health.
• Call us at 1.415.689.3018
Medical Device Info |
|
This product is a medical device.
|
|
| Rx Only. Federal law (USA) restricts this product to sale by or on the order of a licensed practitioner. |
all.health, Inc.
Attn: Algo Medical Functional Working Group
Attn: Algo Medical Functional Working Group
200 Kansas St, Suite 205, San Francisco CA 94103
all.health, Inc.
200 Kansas St, Suite 205, San Francisco CA 94103
t: 1.415.689.3018
[END OF IFU DOCUMENT]