AI Accurately Predicts Time of Death Using Blood Metabolites | News Medical
AI Model Improves Post-Mortem Interval Estimation Through Metabolite Analysis
Artificial intelligence is offering forensic science a more precise tool for determining time of death, a capability crucial for investigations ranging from homicide to missing persons cases. Researchers at Linköping University in Sweden and the Swedish National Board of Forensic Medicine have developed an AI model trained on metabolite data from thousands of blood samples, offering a significant improvement over existing methods.
Currently, estimating the post-mortem interval (PMI) – the time elapsed since death – relies on techniques like measuring body temperature, the onset of rigor mortis, and potassium levels in the vitreous humor of the eye. However, these methods become increasingly unreliable as time passes. The new AI-driven approach addresses this limitation by analyzing changes in metabolites, small molecules produced during metabolic processes, which predictably break down after death.
“Death is a strong biological signal,” explains Rasmus Magnusson, a postdoctoral fellow at the Department of Biomedical Engineering at Linköping University, who led the study published in Nature Communications. “When the body dies, a number of biological processes set in. Organs and tissues begin to break down, leading to changes in small molecules in the blood called metabolites. They are broken down in a predictable way that correlates with how much time has elapsed since the time of death.”
The foundation of this advancement is a unique database compiled by the Swedish National Board of Forensic Medicine (RMV), containing blood samples from over 45,000 autopsies collected over nearly a decade. While traditionally used to identify drugs, pharmaceuticals, and toxins, these samples also contain valuable information about the body’s metabolites. A subset of 4,876 samples with known PMIs were used to train the AI model.
“This represents a gold mine of data,” Magnusson stated. Importantly, the research also demonstrated that a surprisingly small dataset can be effective. “A few hundred individuals are enough to build corresponding models, which makes our method useful even in laboratories worldwide that don’t have access to as much data,” he added.
The researchers found their model could predict the time from death to autopsy with a precision of approximately one day, even for cases where the body had been deceased for up to 13 days. This represents a substantial improvement over existing techniques, particularly in cases where significant time has passed.
The AI doesn’t simply analyze the *presence* of metabolites, but rather the *changes* in their concentrations over time. This nuanced approach allows for a more accurate estimation of the PMI. According to Elin Nyman, a docent in systems biology at IMT, the success of the project was somewhat unexpected. “We knew that many external factors affect body decomposition and were surprised that the signal from the body’s metabolites was so strong when it comes to predicting the post-mortem interval.”
Currently, the model estimates the time *since* death, but doesn’t pinpoint the exact time of day. The research team is now focused on refining the dataset to include more precise time-of-death information, with the goal of developing models that can not only estimate the PMI with greater accuracy but also determine the approximate time of day when death occurred.
“Forensic assessments often involve puzzle-like detective work,” says Carl Söderberg, a forensic pathologist and researcher at RMV. “This new tool gives us better opportunities to assess how long someone has been deceased even when a long time has passed since their death, which is of great importance especially in more complex cases. We’re now working on developing even more accurate models.”
The study was funded by the Swedish Research Council, the foundation Forska utan djurförsök (Research without animal experiments), and the Strategic Research Area in Forensic Sciences at LiU and RMV.