Our platform consists of a wearable sensor synced to an iPhone/app connected to cloud-based server with machine learning algorithm to analyze and report individual physical activity remotely. Following patient’s discharge to home, AiCare platform is deployed to the patient via wrist sensor and iPhone. Physical activity data and trend are displayed on iPhone app. Family members are notified upon negative change in physical activity. The AI questionnaire decides as to whether or not the patients need follow-up at the clinic/hospital.
Our solution predicts the clinical outcome based on the individual physical activity differences during the Active phase (walking, standing, or sitting) vs. Resting phase (lying down. The machine learning algorithm and data analytics of their activities will determine healthy recovery at home vs. deteriorating condition requiring hospitalization or death. Based on 17,000 patient/day data at skilled nursing facility, we achieved ROC curve of 100% sensitivity, 94% specificity, and 100% negative predictive value.
We have demonstrated technical feasibility and clinical utility of the technology, consisting of 4 key components for home health monitoring. This capability will address a frequent dilemma experienced by the family members who are taking care of patients as to whether the patient’s condition is deteriorating or not and when to contact the physicians/hospital. Through accurate measurement of the patient’s physical activity, our technology provides reliable and robust clinical guidance to the family members and patients. The cloud-based analytics consist of the following:
- Integrated Sensor Technology Software Platform provides comprehensive web-based tools to gather physical activity data from the sensors, analyze the big data, and allow easy, user-friendly access of the patient’s data at home.
- Physical Activity Data Analytic Engine collects detailed information and analyze the physical activity data automatically 24/7 with highly intuitive user interface.
- Physical Activity Data Anomaly Detection Algorithm to correlate each data point with big data to detect any anomaly in physical activity during the active and resting phases.
- Cloud-enabled AI Engine and Personalized Behavior Database to collect, analyze, and correlate the information via cloud-enable back-end analytic platform to generate rapid and relevant clinical outcome information.
Advanced analytic platform will track the personalized big data to identify any abnormal physical activity gathered from each individual patient and displayed in user-friendly fashion via iPhone app. Algorithms for Simple Moving Average, Bayesian and Probabilistic analyses are conducted over neural network to track home physical activity during both Active and Resting phases. These changes will not be noticed by the patients but by family members to detect deteriorating condition early to seek immediate intervention or to be reassured in those who are recovering appropriately. This technology will introduce a new set of patient monitoring data by assessing physical activity, which represents the most accurate biological and behavioral representation of the holistic physical condition. This innovative technology will bring the clinically relevant data closer to everyday medical practice and enhance our prognostic capability in caring for our patients at home.
Collaboration is explored with regional hospitals to enable rapid enrollment of discharged and COVID patients. This will demonstrate our ability to deploy our technology immediately to help family members who are monitoring their loved ones at home. Specifically, those patients who are suffering from COVID and cannot enjoy direct contact with the family members and caretakers.