JMIR Publications

ABSTRACT

Background:

As the most commonly occurring mental illness the world over, depression poses significant health and economic burdens to both the individual and community. Different types of depression pose different levels of risk. Individuals who suffer from mild forms of depression may recover without any assistance, or be effectively managed by primary care or family practitioners. Yet other forms of depression are far more severe, and require advanced care provided by certified mental health providers. However, identifying cases of depression that require advanced care may be challenging to primary care providers and healthcare team members whose skill sets run broad rather than deep.

Objective:

We leveraged a comprehensive range of patient-level clinical, behavioral, demographic data, acute and chronic conditions, and past visit history data from a statewide Health Information Exchange (HIE), to build decision models capable of predicting need of advanced care for depression across patients presenting at Eskenazi health, the public safety net health system for Marion county, Indianapolis, Indiana.

Methods:

Comprehensive patient datasets were used to develop Random Forest decision models that predicted need of advanced care for depression across: (a) the overall patient population and (b) various subsets of high-risk patients; patients with a prior diagnosis of depression, patients with a Charlson index of >=1, patients with a Charlson index of >=2, and all unique patients identified across the three above-mentioned high-risk groups.

Results:

The overall patient population consisted of 84,317 adult (age >=18) patients. 6,992 (8.29%) of these patients were in need of advanced care for depression. Decision models for high-risk patient groups yielded ROC scores between 86.31% and 94.43%. The decision model for the overall patient population yielded a comparatively lower ROC score of 78.87%. The variance of optimal sensitivity and specificity for all decision models, as identified using Youden’s J Index, are as follows; Sensitivity = 68.79% to 83.91% ; Specificity = 76.03% to 92.18%.

Conclusions:

This study demonstrates the ability to automate screening for patients in need of advanced care for depression across (a) an overall patient population or (b) various high-risk patient groups using structured and unstructured datasets covering acute and chronic conditions, patient demographics, behaviors and past visit history. Further, these results show considerable potential to enable preventative care, and can be easily integrated into existing clinical workflows to improve access to wrap-around healthcare services. Clinical Trial: None