Computer analytical algorithms analyze images to predict pain

AI algorithms that predict pain recently developed by MIT researchers can help doctors know if their patients are pretending.

Massachusetts Institute of Technology researchers have developed an artificial intelligence (AI) capable of analyzing a photograph and predicting the degree to which the person in the photo is suffering. This AI is called "DeepFaceLIFT" - a machine-learning technique that researchers train with videos describing people who are grimacing or showing signs of discomfort. .

This algorithm will help assess whether the patient is honest about his or her pain.

From these videos, DeepFaceLIFT has learned the smallest details on a person's facial expressions, which can help it estimate the extent of the pain they are suffering when compared to a self. Grade pain score. This algorithm can be recognized regardless of age, sex, skin texture ... and surprisingly accurate results.

This AI project sounds depressing, but it is highly applicable in the real world. Currently, the gold standard for pain assessment is the visual-analog scale (VAS), and this VAS scale is based entirely on self-assessment, so it is subjective. High, depending on the circumstances and high oscillation between different people.

A computer algorithm is unlikely to completely replace this pain scales system, because no doctor dares to tell the patient that I will not treat him because my computer says he is. overwhelmed by his pain! However, it will be a useful tool to help complete the VAS ladder system, making it more objective. It will help assess whether the patient is honest about his or her pain.

Researchers are still working on improving this AI to make it a mobile application that can be used anywhere, anytime, and become an impossible tool. missing for physiotherapists.