Cohort-based nomogram for forensic prediction of SCD: a single-center pilot study

Area of Science:

• Forensic Pathology
• Cardiovascular Pathology
• Biostatistics

Background:

• Sudden cardiac death (SCD) diagnosis is crucial in forensic practice.
• Accurate prediction of SCD, particularly ischemic heart disease-induced SCD (IHD-induced SCD), is challenging.
• Existing diagnostic methods may benefit from quantitative predictive tools.

Purpose of the Study:

• To develop and validate nomograms for predicting SCD probability.
• To specifically develop and validate nomograms for predicting IHD-induced SCD probability.
• To integrate these nomograms into practical forensic casework through web calculators.

Main Methods:

• Utilized a large cohort of 3322 cases, randomly split into training (n=2325) and validation (n=997) groups.
• Employed multivariable logistic regression with LASSO or ridge regression for variable selection.
• Established nomograms based on models with the highest Area Under the Curve (AUC) from Receiver Operating Characteristic (ROC) analysis in the validation cohort.

Main Results:

• SCD prediction nomogram achieved an AUC of 0.751 (training) and 0.735 (validation).
• IHD-induced SCD prediction nomogram achieved an AUC of 0.742 (training) and 0.738 (validation).
• Web calculators were created for easy access and application in routine forensic investigations.

Conclusions:

• Developed and validated practical nomograms for predicting SCD and IHD-induced SCD probability using autopsy variables.
• The nomograms demonstrate effective discrimination and calibration, offering a novel approach to cause of death determination.
• These tools are poised to become valuable assets in routine forensic practice for diagnosing SCD.