Princeton-based ODH Inc., a unit of Japan-based Otsuka Pharmaceuticals, and Princeton University are collaborating on an artificial intelligence program to help health insurers and benefits managers determine individuals’ risk scores.
“The health care industry is just starting to come to grips with the potential of machine learning,” ODH President and CEO Michael Jarjour said. “We see a huge opportunity to push the boundaries by improving machine-learning methodologies so we can better identify underlying behavioral and social factors contributing to individuals’ health conditions and target interventions accordingly.”
A machine-learning program developed by Princeton’s Department of Operations, Research and Financial Engineering can factor in social determinants of health to determine overall risk scores for individual plan participants. That would enable insurance companies to offer follow-up care to patients that might have high risk scores and also allow doctors to make better diagnoses and ensure better outcomes.
ODH’s role is to use Princeton’s technology to help insurers and care coordinators scan large amounts of data without specific inputs or instructions. It also can use pattern recognition software to make predictions about future outcomes based on claims data, electronic health records and social determinant factors such as air quality, employment and crime rates.
“This program will collect data on social determinants of health and factor it in,” said Jarjour. “For example, it may spot that there are no supermarkets in a patient’s area, so the patient is eating fast food all the time. The insurance company can spot that and offer health services that are covered.”
Jarjour and Adam Johnson, ODH’s vice president of product development and operations, noted the health care sector’s increased focus on factoring in social determinants, or population health, when administering care.
Earlier this month, RWJBarnabas Health launched its Impact and Community Investment Department, which will specifically focus on ways in which the health system can improve social conditions in the areas surrounding its hospitals. RWJBarnabas CEO Barry Ostrowsky has said the health system finds social conditions account for up to 60 percent of an individual’s health status.
“Princeton [is] focused on building base models that can be used across industries, so we’re taking that approach and marrying it with our health care industry focus,” Johnson said. “That’s where we’re seeing really good synergies.”
“We have work that we’re trying to do in the machine-learning space around understanding more of the factors involved in what leads to a diagnosis,” he continued. “So there are examples of simple variables that go into diagnosing conditions. What we want to do is use machine learning to identify other environmental factors and social determinants of health, such as income status, education level, housing, alcohol consumption, substance abuse, etc.”
Jarjour said the new tools also can help insurance companies cut down on paperwork.
“We have a colleague who used to work in the insurance department, and she said that she would get aggregate data in the form of spreadsheets, and she would go into a room and lay out all of the spreadsheets and then try to follow the data points across 15 spreadsheets,” he said. “She called it death by spreadsheet. Our service can eliminate that.”
Two years ago, ODH engaged in a pilot program with a national insurance company that was so successful the company signed a long-term contract with the insurer, Jarjour said. He added ODH is in talks with other health insurance companies and is close to signing a contract with the U.S. Department of Defense.
“As you can imagine, there are a lot of issues with post-traumatic stress disorder with soldiers, and we believe that we can offer insights to ensure better outcomes for patients at VA hospitals,” said Jarjour. “With our alliance with Princeton, we now can get out of startup mode and move into growth mode.”