This article was published by Frost and Sullivan Medical Technologies E-Bulletin, Volume 10, Issue 1.
They were tiny, inconsequential, and dwarfed by the enormous giants that walked the earth during the Jurassic era. Waiting for their time to come, mammals fought hard and won the battle of survival, and then emerged to dominate the world for the next 65 million years. Like mammals, the artificial intelligence community worked diligently and determinedly through the ups and downs of the artificial intelligence field. Committed to doing something spectacular – and ignoring the financially rewarding rise of the unintelligent technologies – they patiently persevered in their labs and research centers. And now their patience is paying off as their time has come. Welcome to the dawn of artificial intelligence! The coming decades will belong to this technology as it transforms our world and the greatest impact it will make is in the healthcare industry.
The Two Goals
Simplifying and capturing the formidable complexities of healthcare and zooming in on what can dramatically improve and revolutionize it, we should focus on two fundamental goals: 1) Finding new cures (therapeutic and/or diagnostic), and 2) Applying known cures effectively, efficiently, and for all those who can benefit from them. And artificial intelligence is having an impact on both.
Finding New Cures
One of the most silent and often ignored problems of our times is our stagnation in finding new cures. Unlike climate change or jobs, this issue somehow doesn’t climb to the political consciousness, yet it is one of the most consequential problems of our times. The new drug pipeline appears to be as ailing as the diseases it is trying to heal. Despite a 10-fold increase in investment, the results are miserable (Coller and Califf, 2009). There are staggering failure rates of 97%, even before projects reach the preclinical stage (Sams-Dodd, 2013) and 90% after Phase I, are the industry standards (Biotechnology Innovation Organization (BIO) et al., 2016). The proverbial “Valley of Death” concept captures the disconnect between the upstream and downstream drug discovery process and the “valley” requires complex navigation (Rai et al., 2008). Whether failure is due to toxicity, or efficacy (Sams-Dodd, 2013), or due to cost as a function of time and risk (DiMasi et al., 2009), or other reasons like managerial or organizational issues (Buonansegna et al., 2014), the overriding concern is that the human civilization stands naked and hopeless without the prospect of new cures.
With the advent of artificial intelligence, we can expect to close the gap. Specifically, the solutions are coming in the following areas:
- More efficient and smarter basic science and preclinical models
- Smarter devices for preclinical (pattern recognition etc.)
- Genomics and molecular medicine
- Forensic analysis of clinical trials data (what failed, why)
- Sharing of clinical information to help develop new therapeutic options
- Finding new patterns in existing clinical data
- Enhanced predicative ability to determine toxicity, efficacy etc. at early stages of development
- Integrating various aspects of new drug development such as identified by Mullane et al. (Mullane et al., 2014).
Even cancer, which is not a single disease but potentially hundreds or even thousands of diseases, can be considered as a computational problem that can be solved by artificial intelligence (Tenenbaum and Shrager, 2011).
Making Existing Cures More Efficient and Effective, For All
Now enter the clinical side, where artificial intelligence is improving the current standards of care. Just because we have a cure doesn’t mean it is being applied efficiently and effectively for all those who need it. Artificial Intelligence is now transforming clinical healthcare by:
- Improving the diagnostic speed and accuracy by analyzing data and observing never-before-seen patterns. This includes not only enhancing the ability to save lives by improving the speed and accuracy of diagnostics (for example Sepsis, a major killer), but also by artificial intelligence systems learning the ability to read scans.
- Artificial Intelligence systems are being developed and tested for population health management, patient tracking, condition management, hospital workflow management, advanced analytics, and the list goes on and on.
- Social robots, care bots, and healthcare management bots are being developed to help in providing care, patient monitoring, and doing patient or hospital chores.
- The efficiency of hospitals is being increased by using artificial intelligence for claims management, coding and reimbursement.
- In the future, we can expect healthcare kiosks and freestanding autonomous clinics providing primary care.
- On the behavioral health side, we are observing a tsunami of new solutions providing various behavioral therapies and interventions. This area will greatly improve access and diagnostic consistency across behavioral health.
And this is only the beginning. As a civilization, we must challenge ourselves to conquer disease and suffering. Anyone who is, or has a family member, suffering from a disease like cancer knows that the speed and accuracy of finding cures and timely and effective interventions matter. With artificial intelligence, our hopes stand renewed.
The author will be presenting about the above developments at the 22nd Annual Medical Technologies: A Frost & Sullivan Executive MindXchange.
About AL NAQVI
Al Naqvi is the Executive Director of The Society of Artificial Intelligence in Medicine and Healthcare and the Chief Executive Officer of the American Institute of Artificial Intelligence. He is also Editor-in-Chief of the Artificial Intelligence AI post www.aipost.com.
Formerly, he was the Chief Financial Officer of a major healthcare/hospital system and prior to that a consultant in the drug development industry with a special focus on molecular medicine and nuclear medicine. Prior to that, he was Vice President of a Fortune 500 company and a technology entrepreneur. His doctorate thesis is on Artificial Intelligence Governance and his Machine Learning training is from Stanford University.
Biotechnology Innovation Organization (BIO) et al. (2016) Clinical Development Success Rates 2006-2015. (June), . [online]. Available from: https://www.bio.org/sites/default/files/Clinical Development Success Rates 2006-2015 – BIO, Biomedtracker, Amplion 2016.pdf.
Buonansegna, E. et al. (2014) Pharmaceutical new product development: why do clinical trials fail? R&D Management. [Online] 44 (2), 189–202. [online]. Available from: http://doi.wiley.com/10.1111/radm.12053.
Coller, B. S. & Califf, R. M. (2009) Traversing the valley of death: a guide to assessing prospects for translational success. Science translational medicine. [Online] 1 (10), 10cm9. [online]. Available from: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2879158&tool=pmcentrez&rendertype=abstract.
DiMasi, J. A. et al. (2009) Trends in Risks Associated With New Drug Development: Success Rates for Investigational Drugs. Clinical Pharmacology & Therapeutics. [Online] 87 (3), 272–277. [online]. Available from: http://dx.doi.org/10.1038/clpt.2009.295\npapers3://publication/doi/10.1038/clpt.2009.295.
Mullane, K. et al. (2014) Translational paradigms in pharmacology and drug discovery. Biochemical Pharmacology. [Online] 87 (1), 189–210. [online]. Available from: http://dx.doi.org/10.1016/j.bcp.2013.10.019.
Rai, A. K. et al. (2008) Pathways Across the Valley of Death: Novel Intellectual Property Strategies for Accelerated Drug Discovery. Yale Journal of Health Policy, Law, and Ethics. 81. [online]. Available from: http://www.worldcat.org/oclc/809548427.
Sams-Dodd, F. (2013) Is poor research the cause of the declining productivity of the pharmaceutical industry? An industry in need of a paradigm shift. Drug Discovery Today. [Online] 18 (5–6), 211–217. [online]. Available from: http://dx.doi.org/10.1016/j.drudis.2012.10.010.
Tenenbaum, J. M. & Shrager, J. (2011) Cancer: A Computational Disease that AI Can Cure. AI Magazine. [Online] 32 (2), 14–26. [online]. Available from: http://www.aaai.org/ojs/index.php/aimagazine/article/view/2345.