By Vivek Rajagopal, VP and Head, Advanced Analytics and AI, Narayana Health
Global health care expenditures are expected to continue to rise as spending is projected to increase at an annual rate of 5.4 percent between 2017-2022, from USD $7.724 trillion to USD $10.059 trillion.
At least half of the world’s population still lack comprehensive healthcare services.
The statements above, taken from two different reports indicate that despite spending close to $8 trillion as a society, less than half the world's population has access to healthcare. The healthcare industry is in need of radical transformation to increase coverage and reduce costs without dilution in quality.
On the other side, rapid advancement in cloud computing and bandwidth has woken up a long dormant giant, artificial intelligence (AI). While the concept of AI and several of the algorithms and techniques in use in AI today are fairly old, computing power is increasingly making AI practical and relevant. The ability of AI to drastically transform several industries is now becoming clearer, with the healthcare industry being an important one.
Conceptually, AI, in this context, is about pattern recognition. The algorithms are built around estimating the most accurate generalization of the input dataset and applying it on new/unseen datasets to predict the desired information. The healthcare industry naturally lends itself to AI as a major portion of care delivery is about refining the clinical pathway based on past cases. Analyzing a trove of past data can help in early detection of adverse events, accurate diagnosis and framing treatment plans as well. Under early detection, models can predict several indicators like risk of mortality, morbidity, re-exploration and re-admission, to name a few.
The ability of AI to drastically transform several industries is now becoming clearer, with the healthcare industry being an important one
In diagnostics, convolutional neural network based image analytics models can significantly transform radiology and histopathology by increasing throughput, consistency, and accuracy. Although the early detection mentioned earlier was largely centered around in-hospital incident detection, with increasing availability of affordable wearables, well-being can be continuously monitored at home, analyzed and adverse events can be detected early enough to prevent in-hospital episodes. AI can also help define clinical benchmarks based on local data to estimate outliers more effectively. Periodic re-training of models can also help continuously re-estimate and tune clinical benchmarks. In addition to patient care, AI has applications in clinical workforce training as well. With rapid progress virtual reality (VR) and augmented reality (AR), simulation of several in-hospital scenarios are being built and validated. These advancements can help drastically reduce the training time and vastly improve the quality of training.
As with any solution there are limitations and cautions as well. The quality of data is a major problem the industry is grappling with, in truly tapping the potential of AI. Although a huge volume of data is available digitally, large amounts of it carrying valuable information are unstructured, inconsistent, and incomplete. An AI model is only as good as the quality of underlying data. Organizations developing healthcare AI solutions are now employing dedicated workforces to streamline and clean the data for building meaningful models.
The benefits of AI in reducing costs, increasing coverage and improving the quality of healthcare is very apparent. By automating several repetitive, albeit complex aspects of care delivery, AI can help effectively utilize the very limited healthcare workforce bandwidth by significantly amplifying output and reducing information fatigue for the clinicians. With AI being a software solution, benefits such as scalability, replicability, consistency, and low cost at scale will positively impact the healthcare industry. A major question that tags along with AI adoption is if it is going to replace clinicians. The simple answer is, not in the foreseeable future. More than a century ago, detecting the location of a fracture was a key skill. The invention of the X-ray simplified that to a large extent. Has the X-ray replaced or reduced the importance of orthopedicians or has it expanded the field? AI, in the near future will be a reliable, consistent, and highly accurate clinical assistant available to every healthcare facility in the world. AI has the potential to revolutionize healthcare delivery by drastically increasing coverage along with improvement in quality at a fraction of the proportionate healthcare spend.