Chronolife has pioneered the patented neuromorphic algorithm called HOTS (Hierarchy Of event-based Time Surfaces), a machine learning predictive algorithm capable of continuously analyzing complex data streams on low bandwidth such as smartphones or tablets, and detecting pattern deviations.
This capability enables data extraction and analysis on a wide range of connected devices and objects, for a variety of use cases including smart home, veterinarian care, medical imaging, and so forth, where worsening events can be detected before they occur to protect users’ safety, health and wellbeing, sometimes in critical emergency situations. For example, when integrated with our remote patient monitoring devices, HOTS predictive algorithm is intended to predict cardiac decompensation events and generate alerts to health professionals for medical intervention.
We carried out a first successful test for medical applications that demonstrated HOTS’ effectiveness in detecting atrial fibrillation based on electrocardiogram public data.
Chronolife subsequently carried out a second test on data from 60 patients with sleep apnea, demonstrating once again the ability of HOTS to detect abnormal events such as Sleep Apnea.
HOTS predictive algorithm will continue to get smarter and better with large-scale data training in worsening heart failure (pending clinical trial across 552+ patients in Europe), and others.
| Cardiovascular diseases | ||||
|---|---|---|---|---|
| Signal | Sensitivity (%) | Specificity (%) | Database | |
| Atrial fibrillation | ECG | 100% | 94% | MIT-BIH Arrhythmia Database |
| Extrasystoles | ECG | 95% | 96% | MIT-BIH Arrhythmia Database |
| Respiratory diseases | ||||
| Sleep Apnea | SpO2 Abdominal breath |
96% | 96% | Apnea-ECG Database |
| Sleep Apnea | ECG Abdominal breath |
92% | 92% | Private |
The proliferation of IoT and IoMT has transformed every industry and how people manage their daily lives. At the same time, these connected devices bring about an explosion of data that can yield critical insights for predicting and detecting worsening events before they occur.
The HOTS predictive algorithm is capable of analysing multiparametric data streams continuously, on small computing units such as smartphones or tablets, which greatly increases its compatibility and coverage areas for HOTS to be ported and integrated with a wide range of mobile devices and platforms for local analysis.
Chronolife offers the HOTS as an embedded predictive algorithm that can be easily integrated with any of your devices, platforms… that can collect data. Our team of experts will help you with everything from testing and custom configuration to implementation and ongoing maintenance. We’ll also collaborate with you in training and improving the predictive algorithm based on your unique use cases and data requirements.
