Crystal Bone: Predicting Short-Term Fracture Risk From Electronic Health Records With Deep Learning.

Researchers

Journal

Modalities

Models

Abstract

Fractures due to osteoporosis and low bone mass are common and give rise to significant clinical, personal, and economic burden. Even after a fracture occurs, high fracture risk remains widely underdiagnosed and undertreated. Common fracture risk assessment tools utilize a subset of clinical risk factors for prediction, and often require manual data entry. Furthermore, these tools predict risk over the long term, and do not explicitly provide short-term risk estimates necessary to identify patients likely to experience a fracture in the next 1-2 years.
The goal of this study was to develop and evaluate an algorithmic approach to the identification of patients at risk of fracture in the next 1-2 years. In order to address the aforementioned limitations of current prediction tools, this approach focuses on a short-term timeframe, automated data entry, and the use of longitudinal data to inform the predictions.
Using electronic health record (EHR) data from over 1M patients, we developed Crystal Bone, a method that applies machine learning techniques from Natural Language Processing to the temporal nature of patient histories to generate short-term fracture risk predictions. Similar to how language models predict the next word in a given sentence or the topic of a document, Crystal Bone predicts whether a patient’s future trajectory might contain a fracture event, or whether the “signature” of the patient’s journey is similar to that of a typical future fracture patient.
The proposed models accurately predict fracture risk in the next 1-2 years for patients aged over 50 years (area under the receiver operating characteristics curve [AUROC]=0.81 in a holdout set with 192,590 patients). These algorithms outperform the experimental baselines (AUROC=0.67) and have shown meaningful improvements when compared to a retrospective approximation of human-level performance, such as correctly identifying 9649 of 13,765 (70%) at-risk patients who did not receive any preventative bone-health-related medical interventions from their physicians.
These findings indicate that it is possible to use a patient’s unique medical history as it changes over time to predict the risk of short-term fracture. Validating and applying such a tool within the healthcare system could enable automated and widespread prediction of this risk and may help with identification of patients at very high risk of fracture.