Algorithms power the background processes of modern life. They dictate action on the trading floor, guide motorists through morning traffic and unite unsuspecting soul mates. Soon, they will transform medicine.

Over the past decade, medical researchers and computer scientists have slowly integrated the algorithm into clinical culture. Currently, they are commonly used to run large-scale comparative analysis inside medical record or digital image databases. This is just the beginning.

In the very near future, these enterprising lines of code won’t just analyze clinical data – they will manufacture it.

Desktop diagnosis
In June of 2014, researchers from Oxford University built a computer program capable of diagnosing genetic conditions in children, reported Wired. The program scans 2-D images and then uses a machine learning algorithm to compare the photographs to images of individuals diagnosed with genetic disorders like Down’s syndrome. Over time, it develops what its creators call clinical facial phenotypes, or syndrome-specific image archetypes. The program also organizes scans with similar facial characteristics into groups which is particularly useful for physicians in search of rare, or possibly unidentified genetic conditions.

The development team at Oxford is particularly excited about the program’s uncomplicated workflow.

“Since any normal 2-D image can be analyzed, this approach is available to any clinician worldwide with access to a camera and a computer,” the researchers noted in their study, published in the journal eLife. “This can also reduce the need for patient inconvenience in a clinical setting because a family photo album could provide the required image(s).”

Diagnosis is a popular area of focus for many medical researchers involved in algorithmic studies.  

In 2013, IBM partnered with a number of top hospitals to develop image-scanning technology capable of doling out diagnoses, reported The New York Times. Watson, the company’s artificial intelligence engine, now anchors multiple diagnostic programs. Physicians at Memorial Sloan Kettering Cancer Center, the University of Texas MD Anderson Cancer Center and the Cleveland Clinic use its processing power to develop custom treatment plans for cancer patients. And, in November 2015, doctors at Boston Children’s Hospital leveraged Watson to identify a rare kidney disease, reported Modern Healthcare.      

“Coping with an undiagnosed illness is a tremendous challenge for many of the children and families we see,” Dr. Christopher Walsh, head of the hospital’s genetics team, said in a statement. “Watson can help us ensure we’ve left no stone unturned in our search to diagnose and cure these rare diseases.”

How humans fit into the equation
Some worry that evidence-based machines like Watson could replace the personal touch of human physicians. Others are concerned about more serious issues. Namely, computer errors.

Dr. Robert Watcher, a professor at the University of California, San Francisco Medical School, published a piece on a patient who received the wrong medication as a result of an automated prescription system. The patient, a 16-year-old boy named Pablo Garcia, nearly died.

“The error that nearly killed Pablo Garcia illustrates the double-edged sword of healthcare IT,” wrote Watcher.

According to a study conducted by Health Affairs, most patients think evidence-based care attempts to meet only the minimum quality requirements. Conversely, respondents said they believed most physicians went above and beyond to provide excellent care. 

Surprisingly, even those involved in implementing complex medical technology are skeptical, reported The Wall Street Journal.   

“In medical data, there’s lots of ambiguity and lots of fuzziness,” Dr. John Eng, a radiology professor at the Johns Hopkins School of Medicine, one of IBM’s partners, said in an interview with the paper. “It’s kind of messy data, and I think that’s going to be a limiting factor with what IBM does with Watson.”   

 Despite their immense power, algorithms and the technology they empower still require a human helping hand, for now. 

“The goal is not necessarily to have a human look at the outcome afterwards, but to improve the quality of the classification for the individual,” Gary King, head of Harvard University’s Institute for Quantitative Social Science, said in an interview with The Times. “Often, the two can be way better than the algorithm alone.”