Researchers have used machine learning to predict wellbeing—not only objective physical condition but also subjective overall health—as a function of demographic, socioeconomic and geographic factors.
In the process the team found that socioeconomic variables, especially job dissatisfaction and financial stress, have particular predictive power and can help explain gaps between physical and overall healthiness at the individual level.
Such insights can inform interventions and other care decisions aimed at warding off poor outcomes, the researchers suggest.
The subpopulation on which the team trained and tested their machine learning models comprised U.S. military veterans. The data source was Gallup’s U.S. daily tracking survey from 2014 to 2017.
The study report is set to run in the June edition of Computers in Biology and Medicine.
In clinical settings, the authors write, appropriately weighting socioeconomic variables in predictive models “will provide opportunities to create real-time and personalized feedback to help individuals improve their quality of life.”
They suggest their findings are likely generalizable to vulnerable subgroups of broader populations than U.S. veterans.
The study’s lead author is Christos Makridis, PhD, of Stanford’s Human-centered AI Institute and the National AI Institute at the Department of Veterans Affairs.
The study is available in full for free.