How Artificial Intelligence Lights up the Twilight in the Medical Night

advertisement

Secrets lurking under a microscope are often more daunting than the most complicated cryptography. When the arrangement of cells on pathological sections presents a strange symmetry, human eyes may mistake it for meaningless noise. However, the artificial intelligence algorithm can capture the weak signal before the disease occurs from tens of millions of cell forms-this ability is reshaping the path for us to find "no disease".

IMG_256

The fine folds of extracellular matrix used to be the neglected background color in medical textbooks. The deep learning model trained by the research team of the Federal Institute of Technology in Lausanne, Switzerland, can predict the early risk of myocardial infarction from the arrangement density of collagen fibers by learning 5 million images of myocardial cells. This diagnosis method based on extracellular microenvironment opens the observation window to the eve of the disease.

The noise of metabonomics hides the solution to the puzzle of life. In the organ-like experiment of Francis Crick Institute in Britain, AI model is being trained to identify spatiotemporal anomalies in metabolic debris of cancer cells. When the tumor sample differentiated into the fifth generation in the culture dish, the machine was able to draw a prediction map of tumor evolution from the subtle changes of amino acid metabolism pathway. This dynamic monitoring technology makes the early diagnosis of cancer move from static image diagnosis to real-time tracking.

IMG_256

The axon ends of nerve fibers tremble with the warning of death. The biological simulation system developed by brain like computing team of Oxford University can capture the characteristic waveform of Parkinson's disease in the prodromal stage from the electrophysiological signals of retinal ganglion cells. When the apoptosis of dopaminergic neurons has not exceeded the critical point, the algorithm has been able to gain valuable intervention time for patients.

When we continue to explore along the nano-scale medical territory, we will find that the tentacles of artificial intelligence have penetrated into more secret corners. In the laboratory of the University of Amsterdam in the Netherlands, researchers are training neural networks to identify early signals of mitochondrial diseases caused by mitochondrial DNA mutations. In the past, minor faults in this energy factory had to wait until the organ function was obviously damaged, but now, through in-depth analysis of circulating mitochondrial DNA in the blood, the machine can give an alarm several months before symptoms appear.

What is even more amazing is that in the interdisciplinary research of Tokyo University in Japan, researchers are combining deep learning with single cell sequencing technology to build an "early warning system" of human immune system. Through continuous monitoring of peripheral blood mononuclear cells of thousands of healthy people, AI model has learned to identify the initial response mode of immune system to potential threats. This ability allows doctors to accurately intervene in the inflammatory reaction that is about to get out of control before the full-scale outbreak of infection, even before the patient himself has noticed any abnormality.

IMG_256

When medical diagnosis moves from dominant morphology to recessive molecular dynamics, artificial intelligence is becoming a lamplighter in the medical dark night. Those tiny anomalies that were once classified as "normal variation" are being redefined as quantum signals of diseases. We may be standing at a turning point: from medicine that discovers lesions to medicine that foresees lesions. Every iteration of the algorithm is like lighting a new lighthouse in the endless medical universe, illuminating the health threats that were once hidden in the dark for us. The collection of these technologies will eventually form a new defense matrix for human beings to fight against diseases, so that the tentacles of medicine can touch the earlier end of time and save those lives that have not yet been detected.

WriterWeink