In their bid to be able to predict which patients are likely to fall ill, clinicians are turning to additional sources of data to build algorithms, the Wall Street Journal reports.
While information about patients' medical conditions, prescriptions, and lab test results have been used to single out patients who, for instance, are at risk of being re-admitted to a hospital, some doctors say that adding in consumer data might help refine such algorithmic tools. "So much of what determines a person's health and well-being is independent of medical care," Rishi Sikka, the senior vice president of clinical operations at Advocate Health Care, tells the Journal. Her hospital system developed such an algorithm to predict readmissions based on clinical data, and it reduced the number of readmissions in its first few months of use by 20 percent. The company is now planning on acquiring consumer data to strengthen the algorithm's predictive capabilities.
A few studies, the Journal notes, indicate that nonmedical information, such as data on patients' ability to dress and care for him- or herself, the stability of their living arrangements, and their ability to navigate the healthcare system, can help improve algorithms' performances.