Death report analysis improves suspected overdose predictions with expert explanations
Evidence-Anchored Overdose Death Prediction from Death Investigation Reports
Computation and LanguageArtificial IntelligenceMachine Learning
Summary
Overdose deaths can be hard to confirm quickly because the investigations take time. The authors created a computer method to read death investigation reports and guess if the death was likely caused by an overdose before the official cause is known. Their system also explains its decisions based on important parts of the reports, helping experts understand why it made a prediction. They found that a step-by-step method that first finds important evidence then makes a decision works better than other approaches, and their method beats the current rule-based system by 15%.
What this means in practice
- •For public health officials: Improve early identification of suspected overdose deaths from investigation reports to speed up monitoring and response efforts.
- •For medical examiners: Assist examiners by providing automatic suspect overdose predictions with explanations to support and prioritize investigations.
Authors
Raavi Gupta, Ziqing Yang, Karli R. Hochstatter, Talia Nadel, Nicole D'Anna, Jason K. Graham, Hannah Johnson, Kelsey Opozda, Nabila El-Bassel, Anubhav Jangra, Smaranda Muresan
Abstract
Confirmation of overdose deaths in the United States is often delayed due to the complexities of certifying drug-involved deaths. Death investigation reports represent a rich but underutilized data source that could be used to predict whether a death is a suspected overdose (i.e., a death preliminarily identified as an overdose prior to the final cause and manner of death determination). We present an automated framework that classifies suspected overdoses from death investigation reports, while producing supporting explanations guided by expert knowledge. Through an expert-guided design process, we investigate 1) what type of explanations are best for model prediction performance, and 2) which architecture is better suited for this task: pipeline models that rationalize first and then predict, or self-rationalizing models that jointly predict and rationalize? Our experiments show that a pipeline model with explanations grounded in the input death investigation reports outperforms the other settings. Our best model also outperforms a rule-based system currently in deployment by 15%, while allowing expert interventions for final decisions.