Artificial Intelligence/Machine Learning based augmentation to care

Digital transformation of healthcare now includes big data analytic methods that have opened the door to artificial intelligence (AI) and Machine Learning (ML) solutions.
Typical tasks of ML algorithms include classification, prediction, pattern recognition, and clustering and feature identification. While supervised ML algorithms are trained on labeled structured data, unsupervised ML recognizes patterns in data without pre-specified structure. Deep Learning (DL) algorithms or Neural Networks create complex combinations layers of ML models and thus are able to identify potential predictive values more advanced than the original features. Combine this with Natural Language Processing (NLP) where a computer program can understand human language and now, we have a very intelligent and sophisticated addition to the NLP/DL models. Imagine then a software program (or a robot) that can be driven with this sophisticated predictive model which can be trained to respond or “think” more like human behavior. This leads us into AI prescriptive models which can help patients to manage their health and care teams to learn more about providing more effective care.

Wellpop is working on AI/ML solutions in the following areas:

Identification, Stratification and Risk Scoring

Using multiple data streams from claims, medication, chronic conditions and biometrics, we use clinical markers to stratify patients and provide a risk score on the patient. This risk score can be further used for predictive analysis into patient health trajectories, focused clinical interventions and medical and pharmacy spend.

Patient Engagement- intelligent notifications and care coordination

Focus areas are identified. These can either be a chronic condition (HTN, COPD) or a Medication management Issue (Adherence, Medication protocol), Remote patient monitoring (BP and weight measurements). The ML algorithm analyzes the risk score and can highlight on a dashboard which patients may need outreach and interventions. Furthermore, using AI bots, we can send intelligent reminders, notifications of encouragement or even short questionnaires or make appointments that can personalize the care experience.

Care Management/ Aging at Home

We understand that our daily activities, self-management and education of health and the daily coordination of care are all related to ultimately our health outcomes. Sensor technologies including biometric devices, activity sensors, fall detectors, sleep quality measurements etc. can further assist in understanding health metrics. But making senses of all this data can be overwhelming. AI can assist in creating personalized plan for individual patients and seniors assessing the stage of disease, their awareness of self-care, medication adherence, social determinants of health and lifestyles. Furthermore, AI can monitor for “alert” patterns and without any compromise in personal privacy, share those with a care team member or family member to ensure that a senior with a chronic condition or living alone can be attended to at all times.

Medication Management

It is estimated that 20-30% of prescriptions are not even filled, or picked up at the retail pharmacy. Adverse drug events contribute to a high percentage of visits to the Emergency Department. Pharmacists spend a significant part of their time in medical reconciliations, understanding usage and factors that contribute to medication adherence. ML models can be trained to assist pharmacists in various tasks like medication reconciliation, synchronization, education and patient messaging.