How machine learning can transform medicine

1 March 2019 (Last Updated March 1st, 2019 11:45)
How machine learning can transform medicine

Machine learning is an often-used term that has been promised to do everything from making workers more productive to taking over individuals’ jobs entirely. Frankly, it will likely be many years before anyone should be concerned about being replaced by artificial intelligence (AI) at their job. However, doctors might find AI impinging upon their jobs sooner rather than later.

The medical field has some characteristics that make it an attractive target for machine learning. The high stakes nature of correct disease diagnosis, coupled with over-worked and fatigued doctors, can lead to cases where patients with easily treatable diseases go undiagnosed and suffer greatly from this. Combined with a bottleneck in diagnoses due to the limited number of doctors available and expensive diagnosis costs, machine learning algorithms can appear very attractive for both patients and clinics to implement. If machine learning is able to take over the process of diagnosis, it would be able to process huge numbers of medical diagnoses, freeing up doctors to work directly on saving lives.

Fields that use Optical Coherence Tomography (OCT) such as ophthalmology or X-ray-based imaging (as in cancer or pneumonia) stand to benefit the most from these technologies due to their high volume and large complexity. As medical technology progresses, these scans can become more and more detailed. While increased detail makes the OCT scans more accurate, it also makes them harder for doctors to interpret, as there is more data to process. However, machine learning would not suffer from this issue.

IDx, a company specializing in machine learning-based diagnosis of diabetic retinopathy, has recently caused a huge splash in the medical field. Its machine learning algorithms are able to perform as well as or better than practicing clinicians in detecting a variety of retinal diseases, such as diabetic retinopathy. Where the patient would previously have had to make an appointment and wait a week for the results, now the AI will be able to diagnose patients in as few as five minutes. This allows for many more patients to be processed, and thus potentially treated for this disease. In America, 8,000,000 people suffer from diabetic retinopathy and it is estimated that about 24,000 people go blind due to this disease per year. Clearly, the speedier diagnoses of IDx’s system will allow for many more patients to get tested for this life-changing disease.

One chief consideration of healthcare is always cost. Treatments are evaluated by the amount of good they do versus the cost of the treatment. Therefore, cost-effective treatments are very desirable in today’s cash-strapped healthcare climate. Normally, a doctor-driven diabetic retinopathy diagnosis will cost $130–150, which makes the possibly unnecessary appointment unattractive to many potential patients. However, IDx’s system will only cost patients $30–50, which will allow healthcare systems to reimburse many more procedures and will simultaneously be more attractive to many cost-conscious families as it costs only a fraction of the alternative procedure.

This is only one of many exciting new machine learning algorithms that stand to impact the medical field. It seems likely that machine learning will be able to take over more and more data-intensive medical diagnostics as the technology develops further. This may eventually result in a future where medical diagnoses are done entirely by computers, freeing doctors to focus entirely on treatment.