A team of Bay Area researchers built a deep learning artificial intelligence (AI) model capable of accurately forecasting patient outcomes in rheumatoid arthritis (RA). Their findings were published in JAMA.1
Deep learning, a subset of AI, mimics the workings of the human mind. In these systems, artificial neural networks – algorithms modeled after the brain – learn from large amounts of data.2
The researchers’ deep learning model used electronic health record (EHR) data to predict disease activity in patients with RA at their next clinic visit. They pulled EHR data – including medications, demographics, laboratories, and prior disease activity – on 578 patients from a university hospital and 242 from a public safety net hospital.1
To quantify the performance of their model, the researchers used the area under the receiver operating characteristic curve (AUROC). A composite index score was used to measure disease activity. At the university hospital, the model reached an AUROC of 0.91 in a cohort of 116 patients. At the safety net hospital, the model had an AUROC of 0.74 in a cohort of 117 patients. Baseline prediction using the patients’ most recent disease activity score yielded a statistically random performance in both settings.
“The findings suggest that building accurate models to forecast complex disease outcomes using EHR data is possible and these models can be shared across hospitals with different EHR systems and diverse patient populations,” the authors wrote. “In the future, models built from large pooled patient populations may be the most accurate, giving everyone access to the most robust models trained on the largest and most diverse patient populations possible.”
References
- Norgeot B, Glicksberg BS, Trupin L, et al. Assessment of a deep learning model based on electronic health record data to forecast clinical outcomes in patients with rheumatoid arthritis. JAMA Netw Open. 2019;2(3):e190606.
- Marr B. What is deep learning AI? A simple guide with 8 practical examples. Forbes. October 1, 2018. Accessed June 7, 2019.