If something can’t be measured, it doesn’t exist.  That’s why Haverhill, MA physician Dr. Duncan MacDougall in 1907 put especially sensitive scales beneath the legs of a death bed so he could measure the weight loss at the moment of expiration.  By that means he claimed the departing “soul” weighed about three-quarters of an ounce, or twenty-one grams.

In theory, electroencephalographs and PET scans can measure emotions and perhaps even wisdom. And so on.

Clinical medicine, of course, relies on a thicket of diagnostic measurements.  The normal range blood urea nitrogen in urine is eight to twenty milligrams/deciliter.  Men are supposed to have 200 to 1,200 micrograms/deciliter of testosterone in their blood; women, twenty to sixty micrograms/deciliter.

Hundreds of diagnostic tests exist for screening or monitoring and their indications usually come with a number.  There’s even a formula for how much testing is likely to benefit a patient.  Correlate all those figures and you have the “digital patient,” with an electronic medical record that can feed data into an algorithm for health assessments.

Like it or not, each of us is a living manifestations of a gigantic math problem doctors sometimes can solve.

Genomics and proteomics deploy entirely new genres of measurement.

DNA sequencing can give counts of missing or multiple repeats of “steps” (base pairs) in the double helix “ladder,” while microarray instruments determine the level of gene expression for dozens or hundreds of genes at a time.

Proteomics mainly relies on mass spectroscopy to determine an individual’s “protein profile” regarding a disease state.  This involves shooting a laser beam at a sample target (e.g. blood), scattering its particles to reveal their mass and charge features as a signature.  Then, comparing an individual’s protein profile to others of known health states provides an indication.

With both genomic and proteomic data, artificial intelligence programs do the calculations in the emerging field of “computational medicine.”

Computers are incapable of a friendly “bedside manner,” but they can give doctors some extremely valuable insights into the digital patient who may have a warm smile or a worried look.

Computation can show whether a patient with a primitive blood cancer has acute lymphoblastic leukemia or acute myeloid leukemia, which can be hard to determine with standard pathology methods. The difference is critical because the chemotherapy regimens are different.

It can give the probability after brain tumor resection that no residual malignancy remains.   On a larger scale, running the numbers may be able to show that complex conditions, like Alzheimer’s or vascular disorders, are actually composites of several simpler diseases.

Or, it could unscramble the misdiagnoses of familiar, but vaguely defined diseases, like cerebral palsy or epilepsy, as genetic diseases. And on a larger scale yet, it could reveal the salient health features of tribes, generations of families, religious adherents, workers in hazardous occupations, ethnic populations, or even nations of people.

The algorithm will show that your number is unique, just like everyone else’s.