Non-invasive, rapid disease screening applications utilizing intelligent classification of one or more bio-sensing inputs (i.e. temperature, cough and respiration acoustics, EKG, SpO2, HRV) are critical tools for healthcare decision support. Used as such, these applications can help guide medical personnel in administering clinical diagnostic testing and equip health screeners in protecting overall public health and safety.
The importance of such screening tools has become acute with the recent global COVID-19 pandemic. Integration of effective screening technology will be a key component in daily screening for employers phasing in return-to-work and performing regular hazard assessments for workplace safety.
Screening applications dependent on cloud-based AI suffer network latency that can undermine performance and practical application for mass screening applications such as worker scanning in large plants / facilities, and public screening in hospitals, airports, and sports complexes.
Remote analytics from data center and server algorithms only works when network connectivity and performance supports the transmission of edge sensor data which makes such systems less robust for remote and mobile usage.
Reliance on cloud AI processing of raw sensor data introduces data privacy concerns from transmission of sensitive personal data over the internet.
SensiML edge AI models can analyze rich multi-biosensor and acoustic data locally on smartphone and IoT edge devices for real-time analysis and classification of inputs autonomous from the network.
SensiML has developed an acoustic cough classifier and is working with a leading healthcare provider network, academia, and biosensor device manufacturers to combine large scale datasets of cough acoustic and temperature data to pilot COVID-19 screening applications that can improve upon existing non-contact temperature scanning.
Individuals interested in contributing to our crowd-sourced phase dataset can visit our COVID-19 data collection site to learn more.
Autonomous real-time AI analytics for rapid multi-sensor assessment independent from network and cloud processing.
Fault tolerance with standalone screening algorithms that can execute on the smartphone or embedded IoT sensor device
Data privacy assurance with local processing of sensitive personal sensor data without transmission over public networks or storage and processing in cloud data centers