A Personalized Predictive Framework for Multivariate Clinical Time Series via Adaptive Model Selection

Personalized event prediction for Electronic Health Records

Clinical event sequences consist of hundreds of clinical events that represent records of patient care in time. Developing accurate predictive models of such sequences is of a great importance for supporting a variety of models for interpreting/classifying the current patient condition, or predicting adverse clinical events and outcomes, all aimed to improve patient care. One important challenge of learning predictive models of clinical sequences is their patient-specific variability. Based on underlying clinical conditions, each patient's sequence may consist of different sets of clinical events (observations, lab results, medications, procedures). Hence, simple population-wide models learned from event sequences for many different patients may not accurately predict patient-specific dynamics of event sequences and their differences. To address the problem, we propose and investigate multiple new event sequence prediction models and methods that let us better adjust the prediction for individual patients and their specific conditions. The methods developed in this work pursue refinement of population-wide models to subpopulations, self-adaptation, and a meta-level model switching that is able to adaptively select the model with the best chance to support the immediate prediction. We analyze and test the performance of these models on clinical event sequences of patients in MIMIC-III database.

Time Series Analysis and Prediction of Intracranial Pressure Using Time-Varying Dynamic Linear Models

Intracranial pressure (ICP) monitoring is a key clinical tool in the assessment and treatment of patients in a neuro-intensive care unit (neuro-ICU). As such, a deeper understanding of how an individual patient's ICP can be influenced by therapeutic interventions could improve clinical decision-making. A pilot application of a time-varying dynamic linear model was conducted using the BrainIT dataset, a multi-centre European dataset containing temporaneous treatment and vital-sign recordings. The study included 106 patients with a minimum of 27 h of ICP monitoring. The model was trained on the first 24 h of each patient's ICU stay, and then the next 2 h of ICP was forecast. The algorithm enabled switching between three interventional states: analgesia, osmotic therapy and paralysis, with the inclusion of arterial blood pressure, age and gender as exogenous regressors. The overall median absolute error was 2.98 (2.41-5.24) mmHg calculated using all 106 2-h forecasts. This is a novel technique which shows some promise for forecasting ICP with an adequate accuracy of approximately 3 mmHg. Further optimisation is required for the algorithm to become a usable clinical tool.

Neural Clinical Event Sequence Prediction through Personalized Online Adaptive Learning

Clinical event sequences consist of thousands of clinical events that represent records of patient care in time. Developing accurate prediction models for such sequences is of a great importance for defining representations of a patient state and for improving patient care. One important challenge of learning a good predictive model of clinical sequences is patient-specific variability. Based on underlying clinical complications, each patient's sequence may consist of different sets of clinical events. However, population-based models learned from such sequences may not accurately predict patient-specific dynamics of event sequences. To address the problem, we develop a new adaptive event sequence prediction framework that learns to adjust its prediction for individual patients through an online model update.