The Impact of Deep Learning + Imagery on the Future of Healthcare
Alongside the input modalities of voice, acoustics and sensory, imagery is already having a profound effect on healthcare and is poised to transform how service is imagined and delivered.
As we turn the focus in our current INSIGHTS Special Series to the impacts of imagery in healthcare, we not only recognize profound opportunities from an investment perspective but also for the future of humanity.
Unlike the voice and acoustics modalities in healthcare outlined in the first part of this series, imagery has been used extensively in healthcare for decades in the form of X-rays, ultrasounds, CT scans, and MRIs. All of these methods have been dependent on a laborious process of human evaluation which is both time-consuming and fraught with the potential for errors.
As we now enter the era of machine learning, augmented AI and deep learning in healthcare, the tables have turned as machines now can, for the first time, “see” and read the contents of the images. This greatly accelerates the process of prevention, diagnosis and treatment of ailments ranging from cancer, to diabetes, brain injuries and asthma by combining these technologies to the imagery modality/input.
Further, a study by National Bureau of Economics Research proves an incremental increase in life expectancy with the use of medical imaging and a study done by Harvard researchers concluded that $385 spent on medical imaging saves approximately $3,000 i.e. a hospital day stay. Add to this new innovation and products coming to the healthcare space which incorporate AI/ML/DL, the impact, benefits and potential become exponential.
Until recently, the downside of imagery in healthcare was that the interpretation of medical images was quite limited to a relatively small group of specific experts as the readings were complex and revolved around a variety of parameters, most importantly a core knowledge of the subject.
At the same time, the volume of imaging has been increasing steadily, and there are now over one billion studies per year worldwide, more than half in the U.S. alone. While this is good news for the patients who get the right treatment when they need it, the huge increase in demand for imaging is far outstripping the supply of radiologists, which grew until 2012 and now has slipped back to 1995 levels. [Source: Merritt Hawkins]
Augmented AI and deep learning to the rescue! As we have seen with companies Azafran already has under LOI and in our watchlist, these advances in tech will both help radiologists to do more with less and, even more exciting, as the datasets and references increase, the new frontier of solutions will be able to prevent and diagnose diseases on a scale never before imagined. From an investment potential and benefit to humanity perspective, this is one of those things that gets our team up every day.
As we wrote about extensively in INSIGHTS Issue 15, Augmented AI is at its best performing calculations orders of magnitude faster than humans, then leaving the creativity, synthesis, and imagination to human specialists, in this case radiologists and research scientists. The way radiology AI can most effectively progress for the foreseeable future is by assisting in constrained domains while also emphasizing humans in the loop: the physician as the ultimate decision-maker, with the computer as an advisor.