Revolutionizing Forensics: How AI Pinpoints Time of Death with Precision

Determining the time of death from a decomposing body has long been challenging and subjective, with forensic investigators relying on intuition and environmental observations. Factors such as temperature, humidity, body composition, and the surrounding conditions contribute to the decomposition rate, often leading to varying conclusions from different pathologists. Artificial intelligence is now stepping in to bring objectivity and consistency to this critical aspect of forensic science.

Researchers at Clemson University in South Carolina, led by Katherine Weisensee and Hudson Smith, have developed an AI-powered tool called geoFOR. This innovative system combines a database of information from thousands of death investigations with machine learning to provide the most reliable post-mortem interval (PMI) estimates—the time elapsed since death. GeoFOR is revolutionizing how forensic investigators approach this essential but imprecise task by analyzing patterns across extensive data.

The tool's database includes over 2,500 cases, adding new data weekly. Around 1,800 of these cases are real-world death investigations, while the remainder come from forensic experiments conducted at "body farms" in Texas and Tennessee. These facilities study decomposition under controlled conditions, providing precise PMI data to train the AI model. Photos and environmental details such as temperature, humidity, wind, soil type, and the progression of decay over time are all part of the dataset. This comprehensive input allows geoFOR to identify patterns and make connections that human investigators might overlook.

To use geoFOR, forensic investigators input specifics about a case into an app. Details such as the location of the body enable the AI to factor in local weather conditions. Observations about the deceased, including their body type, evidence of insect activity, and signs of animal scavenging, are added alongside indicators like "purging"—a scenario where fluids ooze from the corpse. Based on these inputs, the AI compares the case to its training data and provides a likely PMI, offering a range of days with an 80% confidence level.

This tool has already demonstrated utility in routine investigations and more complex cases. For instance, a body discovered in a shaded ditch in northern England in 2004 was initially misdated due to cooler-than-expected conditions preserving the remains. GeoFOR, which takes environmental factors into account with far greater precision, could minimize such errors.

Beyond its potential to aid criminal investigations by checking alibis and solving crimes, geoFOR is also valuable for identifying human remains and bringing closure to families. By combining the time of death with the last-known sighting of an individual, the model narrows the timeframe, making it easier to match unidentified bodies with missing person records. In some cases, this process provides bereaved families with a clearer understanding of their loved one's final moments, a vital step in the grieving process.

As forensic experts and investigators worldwide continue to contribute cases and data to the geoFOR database, the tool's accuracy is expected to improve. With AI offering unprecedented insights into the decomposition process, geoFOR represents a significant leap forward in forensic science, combining technological innovation with a profoundly human purpose.