Key result
An LSTM deep learning model predicted intra-operative hypotension 5 minutes before onset with an AUC of 0.83, which improved to 0.87 when the model was trained on patient clusters grouped by mean arterial blood pressure trends.
Why the study?
Intra-operative hypotension is commonly observed following general anesthesia and is associated with life-threatening post-operative complications.
Does training an LSTM deep learning model on phenotypically clustered patient data improve the prediction of intra-operative hypotension 5 minutes in advance compared to a baseline model in patients undergoing colorectal surgery?
Observational (n=604)
No
Does training an LSTM deep learning model on phenotypically clustered patient data improve the prediction of intra-operative hypotension 5 minutes in advance compared to a baseline model in patients undergoing colorectal surgery?
Clustering patients by clinical phenotype or baseline blood pressure trends prior to training deep learning models significantly improves the early prediction of intra-operative hypotension.
This is the authors' abstract. We don't add key points for this paper.
Intra-Operative Hypotension (IOH) is a haemodynamic abnormality that is commonly observed in operating theatres following general anesthesia and associates with life-threatening post-operative complications. Using Long Short Term Memory (LSTM) models applied to Electronic Health Records (EHR) and time-series intra-operative data in 604 patients that underwent colorectal surgery we predicted the instant risk of IOH events within the next five minutes. K-means clustering was used to group patients based on pre-clinical data. As part of a sensitivity analysis, the model was also trained on patients clustered according to Mean artelial Blood Pressure (MBP) time-series trends at the start of the operation using K-means with Dynamic Time Warping. The baseline LSTM model trained on all patients yielded a test set Area Under the Curve (AUC) value of 0.83. In contrast, training the model on smaller sized clusters (grouped by EHR) improved the AUC value (0.85). Similarly, the AUC was increased by 4.8% (0.87) when training the model on clusters grouped by MBP. The encouraging results of the baseline model demonstrate the applicability of the approach in a clinical setting. Furthermore, the increased predictive performance of the model after being trained using a clustering approach first, paves the way for a more personalised patient stratification approach to IOH prediction using clinical data.
May enable proactive OR hypotension alerts; leaves open prospective outcome validation before adoption.
No takes yet. Share an insight, caveat, or question.
Garmendia et al. (2023) conducted an observational in Intra-Operative Hypotension (n=604). LSTM deep learning model trained on clustered patient data vs. Baseline LSTM model trained on all patients was evaluated on Area Under the Curve (AUC) for predicting IOH 5 minutes before onset. An LSTM deep learning model predicted intra-operative hypotension 5 minutes before onset with an AUC of 0.83, which improved to 0.87 when the model was trained on patient clusters grouped by mean arterial blood pressure trends.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: