Systems, methods, and devices for detecting viral respiratory illness in presymptomatic and asymptomatic infected persons

A smart watch for detecting an infectious disease within a person during a non-symptomatic time period of the infectious disease in a non-clinical setting, the smart watch comprising:

• a photoplethysmography (PPG) sensor;
• an accelerometer;
• a user interface (UI)
• a memory;
• and
• at least one processor configured for:
• applying an age, a gender, and a body mass index (BMI) of the person to a multi-variate detection model, wherein the multi-variate detection model includes a principal component analysis (PCA) model and a buffer for storing past data;
• collecting first cardiac data by sampling the PPG sensor at a rate greater than 200 Hertz over a first time period and storing the first cardiac data in a buffer;
• collecting first activity level data by sampling the accelerometer at a rate greater than 10 Hertz over the first time period and storing the first activity level data in the buffer;
• applying the first cardiac data and the first activity level data to the multi-variate detection model;
• applying a first time of day associated with the first time period to the multi-variate detection model;
• applying first body position data of the person associated with the first time period to the multi-variate detection model;
• collecting second cardiac data by sampling the PPG sensor at a rate greater than 200 Hertz over a second time period and storing the second cardiac data in the buffer;
• collecting second activity level data by sampling the accelerometer at a rate greater than 10 Hertz over the second time period and storing the second activity level data in the buffer;
• applying a second time of day associated with the second time period to the multi-variate detection model;
• applying second body position data of the person associated with the second time period to the multi-variate detection model;
• characterizing the first cardiac data and the first activity level data as baseline data including:
• associating the first time period with a non-infected time period of the person;
• and
• determining an upper control limit based on the non-infected time period of the person;
• continuously updating the buffer to remove long term trends via the PCA model;
• determining in near real-time a statistically significant anomaly within the second cardiac data using the multi-variate detection model including the upper control limit, wherein the statistically significant anomaly is associated with the non-symptomatic time period of the infectious disease;
• upon determining the statistically significant anomaly, generating an alert indicative of the non-symptomatic time period of the infectious disease;
• and
• providing the alert during the non-symptomatic time period to the UI, wherein the smart watch provides a non-invasive method for detecting an infectious disease.

A non-transitory computer-readable storage medium, the non-transitory computer-readable storage medium storing instructions to be implemented on a smart watch including at least one processor, the instructions when executed by the at least one processor cause the smart watch to perform a method for detecting an infectious disease within a person during a non-symptomatic time period of the infectious disease in a person in a non-clinical setting, the method comprising:

• applying an age, a gender, and a body mass index (BMI) of the person to a multi-variate detection model, wherein the multi-variate detection model includes a principal component analysis (PCA) model and a buffer for storing past data;
• collecting first cardiac data by sampling a photoplethysmography (PPG) sensor on the smart watch at a rate greater than 200 Hertz over a first time period and storing the first cardiac data in a buffer;
• collecting first activity level data by sampling an accelerometer on the smart watch at a rate greater than 10 Hertz over the first time period and storing the first activity level data in the buffer;
• applying the first cardiac data and the first activity level data to the multi-variate detection model;
• applying a first time of day associated with the first time period to the multi-variate detection model;
• applying first body position data of the person associated with the first time period to the multi-variate detection model;
• collecting second cardiac data by sampling the PPG sensor at a rate greater than 200 Hertz over a second time period and storing the second cardiac data in the buffer;
• collecting second activity level data by sampling the accelerometer at a rate greater than 10 Hertz over the second time period and storing the second activity level data in the buffer;
• applying a second time of day associated with the second time period to the multi-variate detection model;
• applying second body position data of the person associated with the second time period to the multi-variate detection model;
• characterizing the first cardiac data and the first activity level data as baseline data including:
• associating the first time period with a non-infected time period of the person;
• and
• determining an upper control limit based on the non-infected time period of the person;
• continuously updating the buffer to remove long term trends via the PCA model;
• determining in near real-time a statistically significant anomaly within the second cardiac data using the multi-variate detection model including the upper control limit, wherein the statistically significant anomaly is associated with the non-symptomatic time period of the infectious disease;
• upon determining the statistically significant anomaly, generating an alert indicative of the non-symptomatic time period of the infectious disease;
• and
• providing the alert during the non-symptomatic time period to a user interface (UI) on the smart watch, wherein the method is a non-invasive method implemented by the smart watch.

A method implemented on a smart watch for detecting an infectious disease within a person during a non-symptomatic time period of the infectious disease in a non-clinical setting, the method comprising:

• applying an age, a gender, and a body mass index (BMI) of the person to a multi-variate detection model, wherein the multi-variate detection model includes a principal component analysis (PCA) model and a buffer for storing past data;
• collecting first cardiac data by sampling an electrocardiogram (ECG) sensor on the smart watch at a rate greater than 200 Hertz over a first time period and storing the first cardiac data in a buffer;
• collecting first activity level data by sampling an accelerometer on the smart watch at a rate greater than 10 Hertz over the first time period and storing the first activity level data in the buffer;
• applying the first cardiac data and the first activity level data to the multi-variate detection model;
• applying a first time of day associated with the first time period to the multi-variate detection model;
• applying first body position data of the person associated with the first time period to the multi-variate detection model;
• collecting second cardiac data by sampling the ECG sensor at a rate greater than 200 Hertz over a second time period and storing the second cardiac data in the buffer;
• collecting second activity level data by sampling the accelerometer at a rate greater than 10 Hertz over the second time period and storing the second activity level data in the buffer;
• applying a second time of day associated with the second time period to the multi-variate detection model;
• applying second body position data of the person associated with the second time period to the multi-variate detection model;
• characterizing the first cardiac data and the first activity level data as baseline data including:
• associating the first time period with a non-infected time period of the person;
• and
• determining an upper control limit based on the non-infected time period of the person;
• continuously updating the buffer to remove long term trends via the PCA model;
• determining in near real-time a statistically significant anomaly within the second cardiac data using the multi-variate detection model including the upper control limit, wherein the statistically significant anomaly is associated with the non-symptomatic time period of the infectious disease;
• upon determining the statistically significant anomaly, generating an alert indicative of the non-symptomatic time period of the infectious disease;
• and
• providing the alert during the non-symptomatic time period to a user interface (UI) on the smart watch, wherein the method is a non-invasive method implemented by the smart watch.

A non-transitory computer-readable storage medium, the non-transitory computer-readable storage medium storing instructions to be implemented on a smart watch including at least one processor, the instructions when executed by the at least one processor cause the smart watch to perform a method for detecting an infectious disease within a person during a non-symptomatic time period of the infectious disease in a person in a non-clinical setting, the method comprising:

• applying an age, a gender, and a body mass index (BMI) of the person to a multi-variate detection model, wherein the multi-variate detection model includes a principal component analysis (PCA) model and a buffer for storing past data;
• collecting first cardiac data by sampling an electrocardiogram (ECG) sensor on the smart watch at a rate greater than 200 Hertz over a first time period and storing the first cardiac data in a buffer;
• collecting first activity level data by sampling an accelerometer on the smart watch at a rate greater than 10 Hertz over the first time period and storing the first activity level data in the buffer;
• applying the first cardiac data and the first activity level data to the multi-variate detection model;
• applying a first time of day associated with the first time period to the multi-variate detection model;
• applying first body position data of the person associated with the first time period to the multi-variate detection model;
• collecting second cardiac data by sampling the ECG sensor at a rate greater than 200 Hertz over a second time period and storing the second cardiac data in the buffer;
• collecting second activity level data by sampling the accelerometer at a rate greater than 10 Hertz over the second time period and storing the second activity level data in the buffer;
• applying a second time of day associated with the second time period to the multi-variate detection model;
• applying second body position data of the person associated with the second time period to the multi-variate detection model;
• characterizing the first cardiac data and the first activity level data as baseline data including:
• associating the first time period with a non-infected time period of the person;
• and
• determining an upper control limit based on the non-infected time period of the person;
• continuously updating the buffer to remove long term trends via the PCA model;
• determining in near real-time a statistically significant anomaly within the second cardiac data using the multi-variate detection model including the upper control limit, wherein the statistically significant anomaly is associated with the non-symptomatic time period of the infectious disease;
• upon determining the statistically significant anomaly, generating an alert indicative of the non-symptomatic time period of the infectious disease;
• and
• providing the alert during the non-symptomatic time period to a user interface (UI) on the smart watch, wherein the method is a non-invasive method implemented by the smart watch.