This article was first published in 听
Remember Theranos, the company that promised to revolutionize longevity by making health data instantly available from blood collected from a finger prick, allowing for 鈥渁lerts鈥� to make lifestyle changes?听
The tiny amount of blood was introduced into the company鈥檚 鈥淓dison鈥� machine, which would then perform hundreds of tests including blood levels of cholesterol, glucose, hormones and markers for infectious disease. People could then resort to appropriate lifestyle and pharmaceutical interventions to avoid an early demise. But the only demise that happened was that of the company 鈥� because the claims of the Edison鈥檚 ability were fraudulent.
However, monitoring health status and predicting future disease from a drop of blood, which was Theranos鈥� fantasy, has become a reality.听
Swift developments in technology have indeed made possible the detection of future health status from a tiny amount of blood through the application of genomics and proteomics. Steps can then be taken to increase the 鈥渉ealth span,鈥� the period of life spent in good health, free from chronic diseases.
The human body can be thought of as a chemical laboratory with thousands of reactions going on all the time. These reactions essentially constitute life. If we want to interfere with these reactions with hopes of improving health, we first have to identify what they do, be it their role in metabolism or the production of body structures.听
Metabolism refers to the reactions involved in the conversion of energy stored in food into a usable form for cellular processes, as well as the breaking down of food into the building blocks that can then be assembled into proteins, nucleic acids and other important biomolecules. Obviously, an understanding of what is going on in the body at a given time is required for making interventions with hopes of improved health. But how do we get a glimpse into the workings of the body?
In the 17th听century, Antonie van Leeuwenhoek produced a microscope that allowed him to see objects far smaller than anyone before him. He described little 鈥渁nimalcules鈥� frolicking in water and saliva. They were later identified as bacteria, leading Louis Pasteur to formulate the germ theory of disease.听
In 1816, French physician Ren茅 Laennec needed to listen to a patient鈥檚 heart, which in those days was done by placing an ear on the chest. But in this case, the female patient鈥檚 physical attributes made that seem impossible. Recalling how he had seen two boys talking to each other through a hollow log, he rolled a piece of paper into a tight cylinder and placed one end on the patient鈥檚 chest and the other to his ear. He had invented the stethoscope.
In 1895, Wilhelm Roentgen discovered X-rays, which led to the non-invasive visualization of the skeletal system. A year later, Italian physician Scipione Riva-Rocci invented the inflatable arm cuff to measure blood pressure. Then in 1902, Dutch physiologist Willem Einthoven developed an electrocardiogram that required patients to put their arms and legs into buckets of salt water.听
MRI and CT scans were introduced in the 1970s, making possible detailed visualization of the body鈥檚 insides. While these instrumental methods were being developed, chemists were identifying proteins and nucleic acids, and physiologists, biologists and geneticists were working out how DNA was the 鈥渂lueprint鈥� for an organism鈥檚 life.听
The invention of X-ray crystallography, the electron microscope, nuclear magnetic resonance (NMR) and mass spectroscopy allowed for the detailed identification of molecule structure, which in turn led to the elucidation of many biochemical reactions, including those involved in metabolism.听
Human intelligence has allowed all these developments to be put together into a pretty good understanding of how the body works.
Now, enter artificial intelligence. This is the use of computers to perform tasks that normally require human intelligence. But computers can do this much, much faster. Furthermore, when provided with data, they can discover patterns that humans cannot see and can make predictions based on the data. For example, 鈥渕achine learning鈥� allows computers to detect spam emails. The machine notices certain patterns in words, senders and links in emails that people have marked as spam and recognizes these in future emails that then are moved into a spam folder. This sort of machine learning has great potential in medicine.
In the UK, the Biobank study recruited half a million people between 2006-2010 who were asked to give blood samples and provide information about their diet, exercise and smoking habits.听
Over the past two decades, their health status has been monitored and their original blood samples have been subjected to proteomics, the study of all the proteins in an organism. This is complex technology, but methods are available to identify the number of different proteins in a sample of blood. The computer is then told which patients eventually suffered from a specific disease and is instructed to look for patterns of proteins in the blood samples that are linked with each disease.听
The machine remembers these patterns and can identify them in the blood sample taken from a person who does not yet have a disease and can predict whether they are likely to develop the disease later. This information can be used to take preventive measures either through altered lifestyle or pharmacological intervention.
Another area where machine learning is destined to become very important is radiology. Feeding millions of X-rays or other scans from patients who have been diagnosed with a disease into a computerized system allows the computer to learn patterns of images associated with a condition. Since it has seen millions of images, the computer can detect patterns that the human eye would never have detected and is more likely to make a correct diagnosis.
In the future, proteomics and genomics (the study of how genes function and vary) may even be used to determine whether a patient will respond to a specific therapy. The computer can be fed proteomic and genomic information as determined from a patient鈥檚 blood sample and be told whether the patient responded to a certain therapy. After being given information from many patients, the machine learns to recognize patterns and predict which therapies are likely to work for a newly diagnosed patient.
Machine learning will also lead to the identifying of people at risk of Alzheimer鈥檚 or Parkinson鈥檚, perhaps decades before the disease appears. This brings up the question about whether one would want to know this, given that there is not much that can be done for prevention. On the other hand, one would want to know about the risk of heart disease because there are preventive lifestyle factors.听
It is amazing how computer science and medicine have partnered. When I took my first computer course, we were using punch cards to program computers and I would never have imagined that one day these machines would diagnose and manage disease.听
As for Theranos, the company executives were convicted of fraud and received jail sentences.
