Human postmortem interval estimation from vitreous potassium: an analysis of original data from six different studies

Area of Science:

• Forensic Pathology
• Toxicology
• Biostatistics

Background:

• Estimating postmortem interval (PMI) using vitreous humor potassium levels (KV) is a long-standing forensic challenge.
• Previous studies show inconsistent linear or piecewise-linear models for KV-PMI, often failing to account for influencing factors like age, urea nitrogen, temperature, and illness.
• Standard linear models assume linearity and constant variance, assumptions not supported by existing KV-PMI data, leading to unreliable estimates.

Purpose of the Study:

• To reanalyze existing data from multiple studies to develop a more robust method for estimating postmortem interval (PMI) from vitreous humor potassium (KV) levels.
• To address the limitations of traditional linear models by accommodating non-linearities and changing variability in the KV-PMI relationship.
• To improve the precision and reliability of PMI estimations across a wider range of cases.

Main Methods:

• Reanalysis of original data from six studies, totaling 790 cases.
• Application of a local regression model (loess smooth curve) fitted separately to each study's data to capture local, non-linear KV-PMI relationships.
• Combination of data from all studies to generate a single loess curve with 95% confidence bands for inverse prediction of PMI.

Main Results:

• The relationship between KV and PMI is demonstrably non-linear with unstable residual variability, invalidating simple linear model assumptions.
• The developed loess model effectively accommodates these non-linearities and changing variabilities, providing a more accurate representation of the KV-PMI relationship.
• Inverse prediction using the combined loess curve yields more precise PMI estimates across the entire KV and PMI range compared to single-study analyses, though reliability decreases as KV increases.

Conclusions:

• A local regression (loess) approach offers a superior method for estimating postmortem interval from vitreous humor potassium levels, overcoming the limitations of traditional linear models.
• The combined loess model provides more reliable and precise PMI estimates by accounting for the complex, non-linear nature of the KV-PMI relationship and its inherent variability.
• This advanced modeling technique enhances forensic pathology's ability to determine time since death, supported by cross-validation for predictive performance.