I was somewhat surprised by a recent article suggesting that algorithms themselves will constitute a new type of medical test. The short article cited the work of Predictive Medical Technologies as a basis for this claim. The company software generates health predictions based on previous clinical data for ICU patients (see: Algorithms are the new medical tests; How data and algorithms help doctors make use of real-time data). Below is an excerpt from the article:
Predictive Medical Technologies claims that it can use real-time, intensive care unit (ICU) monitoring data to predict clinical events like cardiac arrest up to 24 hours ahead of time. Effectively, the startup's algorithms are new types of medical tests that an ICU doctor can take into consideration when deciding on a course of treatment. Predictive Medical Systems is based in the University of Utah's medical accelerator, which is attached to a hospital. The system will soon be tested on a trial basis with real patients and ICU physicians.
I need to give more thought to the idea that the use of an algorithm based on previous data constitutes a medical test. Here's a definition for a medical test from the Wikipedia:
Some medical tests are parts of a simple physical examination which require only simple tools in the hands of a skilled practitioner, and can be performed in an office environment. Some other tests require elaborate equipment used by medical technologists or the use of a sterile operating theatre environment. Some tests require samples of tissue or body fluids to be sent off to a pathology lab for further analysis. Some simple chemical tests, such as urine pH, can be measured directly in the doctor's office....The validity of diagnostic test results produced in each laboratory is entirely dependent on the measures employed before, during, and after each assay. Consistency in the production of good results requires an overall program that includes quality assurance, quality control, and quality assessment.
Within clinical pathology and anatomic pathology, we are accustomed to using computer algorithms as components of tests. For example, IVDMIAs/LDTs require the use of an interpretive algorithm to generate a test result based on data generated in a parallel fashion as part of the same test. Similarly, many digital pathology reports are based on interpretive algorithms. However and in all of these cases, a tissue or blood sample is first submitted to the laboratories. The tissue/blood is analyzed and this information is then acted upon by the algorithm in order to generate the final test result.
I believe that we are now entering an era in which massive clinical databases will be generated in connection with the care of many patients. As proof of this idea, you need think only about whole genome scanning for a patient which will soon become economically feasible. It is certainly not far-fetched to assume that it will be both practical and appropriate to periodically activate algorithms to analyze one's entire medical record, particularly previously reported lab tests. The goal will be to generate predictions about the individual's current and future health status. Such episodes of "algorithmic medical record analysis" would certainly fall under most current definitions of a medical test.