Mar 22, 2019
Artificial Intelligence (AI) has been steadily transforming the healthcare landscape. From faster drug discovery, preclinical and clinical development, precision medicines, robotic surgeons to digital health consultations, chatbots and wearable sensors, the healthcare sector is seeing multiple applications for AI.
By 2021, the AI-enabled healthcare industry is projected to grow to $6.6 billion USD (a CAGR of 40 percent). Beyond personalized patient care the focus with AI technology has also been to decrease the costs of operations across the healthcare sector, faster drug discovery and error-free, efficient & secure clinical trials process.
In this article, we briefly explore five aspects of healthcare where AI is being implemented to dramatically improve the processes and helping in faster drug discovery, build efficient R&D capabilities, deliver personalized care and maintain data security.
Pharmaceuticals are a fairly new avenue for AI in the healthcare industry, with the potential to significantly disrupt the process that companies follow for drug R&D and make their way into everyday medicine cabinets. The average time for a drug to go from the lab to the patient is 12 years (CBRA). Of the drugs that go from preclinical testing to actual human trials, only 5 out of 5,000 (1/10th of a percent) ever make it through successfully, and, even then, only one is approved. Numbers from Tufts Center for the Study of Drug Development show that average costs for new drug development are $2.6 billion.
There is a clear need for clinical research organizations to determine if new, improved methodologies using AI can be discovered to speed up this process in order to put necessary medications in the hands of those that need them. So how can AI actually help with the research, discovery and development of more life-saving medication with fewer hurdles and fewer costs?
Berg Health, a US-based biopharma company, takes a patient’s biological information and uses AI technology to highlight why some people are able to overcome diseases based on a patient’s genetic markers and the environments in which they live. This data is then compiled to propose more efficient treatments and suggest improvements for future treatment, which helps “in the discovery and development of drugs, diagnostics and healthcare applications.”
Atomwise conducted a search of existing drugs with the intention of redesigning them to treat Ebola, without having to start from scratch to find a treatment. From this search, Atomwise found two drugs that fit the criteria to reduce Ebola infectivity using AI technology; as a result, they were able to find a possible avenue within 24 hours.
AI can be implemented in the complex and ever-evolving realm of biosciences, and there are many big-name players and startups stepping up to design cognitive computing to address healthcare needs and the widespread adoption of AI-enabled life science.
One particular avenue would be research and development of modelling and extrapolation from the findings of the vast amounts of health data we’ve collected over the past few decades. Better models allow for better hypotheses and more refined research, leading to more detailed genome profiling, more effective advances in medical devices, better training of practitioners, and more personalized care methods.
Google’s approach to AI in health-tech is its DeepMind Health project, which combines ML and neuroscience research to create powerful learning algorithms that mirror the neural networks of the human brain. This project has brought together world-class researchers, clinicians, patients, and technology experts to solve the healthcare problems we see today to find solutions in the very near tomorrow.
Saama has been implementing a Life Science Analytics Cloud (LSAC) in order to meet the challenges faced with drug trial planning and feasibility, preclinical operations, and contingency plans for adverse drug reactions (their pilot program ran throughout 2018 with several pharmaceutical partners).
Mainstream attention of AI medical devices has been primarily focused on wearables and sensors in the health and wellness industry, including Apple Watch, Fitbit, Garmin monitors, and apps on our mobile devices that track activities, activity levels, heart rate, and sleeping hours.
More cutting edge AI technologies, however, are focusing on actual medical care and improvements in the personalization and quality of healthcare for current and future generations.
The AiCure app is a real-time monitoring solution that confirms whether or not a patient has taken their medications and if they’re taking them at the correct intervals, a feature that is particularly useful for patients who often forget their medication or who go against a doctor’s advice. The app uses a patient’s mobile camera or webcam and AI-enabled technology to confirm that the dose and time taken are correct, supporting the management of their own health.
Face2gene is a search and reference solution that scans a patient’s face and references that information against a database to spot signs of possible disorders. Another example is Remidio, which has been successfully used via a patient’s mobile device to offer a diagnosis for diabetes simply by analyzing photographs of a user’s eye.
Medical companies in the past have prioritized working with products that use historic, evidence-based healthcare. However, as we enter the 4th Industrial Revolution, companies are evolving their solutions and products using AI, VR/AR, and robotics to deliver outcome-based, preventative care.
Using these advances in technology, practitioners are able to more accurately assess “down to the familial and individual level, which one day may even be able to predict and thereby prevent disease.”
We’re seeing real-life advances in fields such as mammograms, where AI tech is able to review and read mammograms “30 times faster with 99% accuracy” and drastically cut down the number of unnecessary biopsies.
Other examples include Deep Genomics, which uses massives sets of genetic data and patient records to determine patterns of diseases and mutations that help doctors discover what happens in a cell when DNA has been altered through natural or therapeutic methods, and Human Longevity, a genome sequencing scan that offers patients an incredibly detailed exam with the added functionality of early-stage cancer or heart disease detection.
Massive amounts of data have been generated in the healthcare industry, from patient histories and medical treatment records to the recent stream of data from wearables in the fitness world.
Good quality data and analytics have often been priced so prohibitively due to the tremendous time and effort of curating what is truly useful and relevant. In addition, the majority of this data (nearly 80%) remains outside of a database or other searchable data structure.
An estimated 4 zettabytes of health-related information was generated in 2013 (4 trillion gigabytes); some are projecting the volume of data will increase tenfold by the year 2020…up to mind-numbing yottabyte proportions.
Researchers and practitioners in the past have been limited to the data they personally know or what their organizations own on (often) archaic systems. Using basic search engines such as Google doesn’t provide the detailed, relevant data required because these algorithms aren’t designed for the intricacy of life sciences and medical research.
A human-centric industry, healthcare is riddled with errors and potential fraud, making the implementation of AI applications ever more critical to protect sensitive data and prevent the exploitation of patients.
Cybersecurity in medicine alone is expected to be a $2B industry by 2021, with more enhanced ways to protect patient data and treatment histories. Experts have estimated that roughly $17 billion a year could be saved by tightening and improving existing security/cybersecurity measures with AI…an area that has traditionally relied on manual and time-intensive processes.
An increasing number of hospitals are being hacked, as many of their devices and systems are connected to the Internet and open to the outside, where they can be hacked. CyberMDX had recently discovered a vulnerability in their syringe pump that allowed hackers to control the device and give patients lethal doses of their medication. In this instance, companies, hospitals, and “advanced cybersecurity solutions could use machine learning” to more quickly understand weak points and detect unusual or suspicious activity to prevent these attacks in the future.
As decision makers, you need “the most current, relevant, and contextual data at your disposal to make the best decisions…AI is making it possible to crawl the endless sources of information out there and provide real-time analytics for [your] organizations.” (Future of Everything)
We have seen a significant increase in long-term illnesses, chronic diseases, and an aging population, rising costs of drug discovery – all factors that contribute to rising costs and workload on the healthcare, pharma and biosciences sector. Many companies, however, have been focused on solutions that only meet the needs we currently have, not the ones we will experience in the future…and this neglects to accommodate for our limited resources and the necessary shift from short-term institutionalized care to longer-term complex care requirements.
While regulations and compliance standards often hamper or inhibit the adoption of technology in the healthcare industry, “the good news is that many of the latest regulatory requirements are compatible with AI exploration.”
Executives and decision makers from pharmaceutical companies to hospitals to clinical research organizations can prepare for a compliance-focused approach with AI-enabled technology. They can do this through industry standards to adhere to; careful consideration of drawbacks and possible solutions of AI; training and education on AI technology for their teams and practitioners; transparent communication with the public about both the benefits and risks of AI-enabled medical technology; and gradually driving these innovations within their own organizations in order to measure their success, affordability, and effectiveness for future medical practices.