The power of our social networks in driving behavior change is well known, especially in terms of our health and wellbeing. A recent study from the Medical University of South Carolina highlights not only the power of our network, but also how this can be enhanced when combined with AI.
The researchers utilized natural language processing to help identify instances of social isolation in the medical records of patients so that remedial interventions can be implemented. It was capable of accurately detecting such individuals 90% of the time.
The findings are important as social isolation is one of the most important determinants of health, alongside things such as education, income and marital status. These social factors have been shown to be as influential as physical aspects, such as blood pressure.
“We know from careful evidence that social determinants are important to health care and health outcomes,” the authors say. “Social isolation is a really important social determinant because it reflects the extent to which people perceive they have a high level of connectedness and support.”
Social determinants of health
There is a growing call for social determinants of health to be documented in our medical records, but busy doctors don’t always have the ability to do so. What’s more, many medical record systems lack the ability to enter such social determinants as coded data. This makes it difficult to document the conversations that are inevitably happening, with the only place to find such information being the clinical notes for that patient.
This free-form documentation presents a challenge for computer systems that attempt to make sense of large data sets. This was emphasized by this research, as the system combed through over 150,000 documents spread across 55,516 clinical notes for over 3,000 patients, all in just a handful of seconds.
This data was then used to train the NLP before testing its capabilities on a fresh set of documentation from over 1,000 new patients. As before, it was able to accurately detect those who were socially isolated 90% of the time.
“It’s pretty darned accurate,” the researchers explain. “It performed well, but the problem remains that some physicians do not comment on these issues and so don’t leave a trail for NLP to follow.”
A strong data trail
The team believe that AI can be of assistance here too, as it could help to search for socially isolated individuals based upon clearly identified traits. They believe this could help to spot such people even if there is no explicit mention of social isolation in the medical notes.
The team also believe that their NLP-driven approach could be applied equally successfully to a range of other social determinants of health, especially if they’re not ones that can be codified in the medical records. For instance, they’re already using such an approach to spot patients with financial security and suffering from alcohol abuse.
“Sometimes physicians focus excessively on the ‘medical’ problems and don’t pay enough attention to the context that people live in and the social aspects that influence their health,” the researchers conclude. “Our study once again highlights the importance of knowing this information in order to provide patients our very best care.”
