Interest in artificial intelligence continues to explode across every industry, but few areas offer more opportunities for drastic improvement of human life than the application of machine learning and AI in healthcare and the medical field.
What does AI mean in medical terms?
Let’s begin first with a definition. AI in healthcare and medicine means using data more effectively through machine learning algorithms to produce positive patient outcomes.
The sheer amount of data created through IoT-enabled devices, the electronic medical record (EMR), and ever-expanding quantities of genetic data has made possible a large number of applications of artificial intelligence in healthcare. Check out the Harvard Business Review ranking of the potential value that these applications could bring to the healthcare industry.
The underlying value of artificial intelligence is to enhance human decision-making and automate processes that are time- or resource-intensive for humans to perform. This promise relies on data – capturing it, analyzing, and using it to provide precise, data-driven answers to critical questions.
Prediction forms a core component of what healthcare professionals do every day. AI can operate as a fast, accurate, and in the long run, cost-effective method to assist human experience and intuition through predictive analytics. AI is not meant to replace doctors, but rather empower healthcare professionals by adding a data-driven context that delivers the right information at the right time, allowing them to make more informed decisions. With more educated decisions, we can be proactive in our care rather than reactionary.
Healthcare applications that leverage artificial intelligence could be used to make more accurate diagnoses, identify at-risk populations, manage and assign administrative resources, forecast the potential value of research projects, and better understand how patients will respond to medicines and treatment protocols.
AI will empower doctors to treat patients more efficiently, even remotely. It suggests exciting food for thought: the developing world may be able to leapfrog the developed world in healthcare delivery. Regions like Africa could skip investment in physical healthcare infrastructure the same it skipped building a robust telecom infrastructure in favor of mobile networks.
AI will unlock potential across many areas of human health. This article will explore 10 specific use cases for AI in healthcare and various medical fields.
What are the most promising AI applications in healthcare?
1. AI-assisted image analysis for radiology
Recent advances in computer vision are set to revolutionize the field of medical imaging, which in turn extends across many different healthcare functions.
In radiology, doctors analyze image sets gathered from, e.g., Computed Tomography (CT) scans, Magnetic Resonance Imaging (MRI), ultrasounds, PET scans, and mammography.
Doctors performed the first CT scan of a human brain in 1971. Harvard Medical School reports that today over 80 million CT scans are performed each year. That’s a lot of pictures to examine by hand, requiring immense amounts of hospital resources to detect diseases, much less detect them in real-time.
AI-assisted imaging technologies expand the ability to analyze these images through pattern recognition. They can help doctors by highlighting certain image features, identify early predictors of cancer, prioritize cases and cut down on the volume of labor required to perform accurate diagnoses.
Training machine learning models on mountains of imaging data optimizes them to detect microscopic anomalies and inconsistencies that indicate the presence of ailments.
CureMetrix is a San-Diego based company that uses advanced machine learning, natural language understanding (NLU), and computer vision technology to assist radiologists in analyzing mammograms for cancer detection.
The company’s use of complex neural networks and disparate sources of training data, including doctor reports and a large database of mammograms, resulted in successful detection of breast cancer up to 6 years earlier than human doctors and enabled a 70% reduction in the number of false positives compared to other solutions currently in the market.
CureMetrix produced such high efficacy in mammogram analysis that the FDA created a new code for their cmTriage (™) platform breast cancer detection driving others to use them as a predicate.
The success of their solution has drawn the attention of some of the most renowned healthcare institutions in the world including MD Anderson, the Mayo Clinic, and Dasa from Brazil.
2. Predictive analytics for hospital resource optimization
Another application of AI in healthcare relates to administrative supply and demand of hospital resources. Crunching vast quantities of data surrounding in-patient volume, types of treatments being administered in real-time, and forecasting future demand can help hospitals allocate their limited budget effectively.
A survey by Health Information and Management Systems Society indicates that 44% of hospitals surveyed are planning on implementing AI into their workflows. Over 80% believed that traditional ERP systems will be entirely replaced by AI and machine learning.
For example, AI-enabled predictive analytics allows hospitals to predict when a flu epidemic may strike a certain location. Administrators can then plan for the number of ICU beds, flu shots, ventilator equipment, and the number of hospital staff necessary to respond effectively.
Artificial intelligence systems can assist human experts by automatically ranking the severity of an impending outbreak by analyzing population-level disease data and automating a hospital’s preparation process on the administrative side.
Check out our AI readiness scorecard to see if your organization is ready to implement artificial intelligence into your resource planning.
3. Improving pathologists’ ability to diagnose tissue samples
The application of AI in pathology is still in its infancy relative to other medical fields. However, pathologists’ analysis of images is well suited for enhancement through machine learning algorithms.
Pathologists spend their days looking through microscopes, analyzing hundreds of slides containing tissue samples. Tiny differences between series of slides often provide the key that unlocks an accurate diagnosis.
Dr. Kelly Bethel, Oncologic Pathologist at Scripps MD Anderson in La Jolla, CA, said “Most medical specialties rely on accurate diagnoses from pathology. Pathology interpretations chart the course for treatment for surgery, dermatology, hematology/oncology, obstetrics/gynecology, nephrology, and urology, among others. AI systems that can improve accuracy or efficiency in pathology will have ripple effects that spread from the individual patient throughout the entire hospital system. Getting that correct influences the entire healthcare system.”
Machine vision techniques for classification, sorting, and differential analysis of tissue samples have come a long way in the last few years.
A project from Memorial Sloan-Kettering analyzed a massive collection of pathology slides to develop algorithms that can identify areas of interest in tissue samples. Algorithms like these will help group samples into categories so that pathologists can quickly and easily identify critical alterations that can then be used to issue a tissue diagnosis.
Pathology is a highly specialized field with a limited number of trained professionals. Any efficiency increases will provide far-reaching benefits, especially when so many other fields depend on accurate results. We expect that machine learning algorithms for image processing will soon emerge as critical decision support systems for pathologists.
4. Guiding population-level disease prevention
Leveraging predictive analytics more consistently across the entire population offers the possibility of shifting the primary point of care away from treatment after-the-fact and more towards disease prevention.
Understanding which individuals are at greatest risk for disease contraction and offering preventative guidance can reduce the likelihood that these patients ever need to enter the healthcare system.
AI can reduce the cost burden to both the patient and the overall healthcare system, which grows more and more constrained as large swathes of the baby boomer generation approach retirement. For both parties, a dollar saved is much more valuable than a dollar earned.
Check out this article by MIT for more info on AI in population-level health control.
5. Delivering individualized precision medicine with AI
The era of precision medicine is here, brought into existence by the confluence of bioinformatics, genomics, electronic medical records, and advances in machine learning.
Physiology, biochemistry, and genetic characteristics vary widely among individuals. Doctors now have the technology to accurately reflect these individualized differences across the entire medical landscape.
Machine learning analysis of genetic data across vast populations of individuals will enable individualized diet recommendations and drug treatment optimizations that increase the probability of successful therapy for diseases of all kinds.
AI-optimized recommendations for treatments will reduce the likelihood of patients suffering from unwanted consequences when prescribing medication. Avoiding a shotgun approach to drug prescriptions will prevent unanticipated drug interactions and improve patient outcomes. Over 25 different drugs have been approved by the FDA this year that target individual genetic sequences.
CureMatch is a leading innovator that addresses the problem of the complexity of cancer. Genomic sequencing is a powerful technology that gives unparalleled insights into the exact DNA abnormalities that are driving an individual’s cancer. But cancers are tremendously complicated because there are hundreds of thousands of potential genetic abnormalities and millions of possible drug combinations. Furthermore, each person’s cancer is different.
Understanding the massive amount of data produced each time a cancer sample is sequenced is an enormous challenge.
CureMatch uses advanced computer algorithms to digest millions of data points derived from individual genomic profiling and then provides oncologists with advanced treatment decision support.
The company’s platform uses a three-step process to analyze patient genomics, identify key individual genetic markers, and rank potential treatment options by their predicted impact on the abnormalities in that patient’s tumor. The platform also assesses whether or not an individual would be a good candidate for certain clinical trials based on their genetic makeup.
6. Machine learning for drug discovery
Pharma brands spend billions of dollars per year on failed drug discovery ventures. The scale, complexity, and high probability of failure of the drug discovery process hamper innovation and, ultimately, increases drug prices for the average person.
AI in pharmaceutical research offers the possibility of generating infinite numbers of novel proteins, drugs, and molecules that machine learning algorithms can sort, prioritize and analyze to determine which ones are viable candidates to move forward through the process.
Large drug companies like Pfizer and Johnson & Johnson already employ large data science teams to analyze molecular models and project chemical interactions via machine learning algorithms.
Successful AI applications in drug discovery are now possible due to the visibility unlocked by genomics into how variability between individuals’ genetic sequences (SNPs) impacts how patients will react to novel drugs.
Generating these molecular combinations on a massive scale will mean huge gains in efficiency for drug companies. This efficiency will translate to lower prices, higher quality of care, and greater impact for the average individual.
7. AI-Enabled clinical decision support systems (CDS)
Another impactful application of AI in healthcare and the medical field is in clinical decision support. Healthcare professionals make recommendations based on a lifetime of medical training and experience. Often it is difficult for machines to mirror this type of performance accurately due to the sheer volume of experience-based inferential information that goes into a seemingly simple diagnosis.
The fundamental promise of clinical decision support systems is to give healthcare professionals the ability to make personalized, valuable, and effective decisions that result in positive outcomes for patients.
One company set to revolutionize the Clinical Decision Support space is Saperi Systems. The company builds information architectures that hospitals can use to store all of the critical domain knowledge that neural networks need to provide predictive value.
The platform organizes patient medical history, care preferences, allergies, genetic information, domain-specific knowledge such as medical literature and treatment protocols, and any other category of information that a doctor may use to make a diagnosis.
The system allows neural networks to reason over that body of knowledge in the context of each patient. The result? It spits out an individualized ‘decision recommendation’ to the healthcare professional that balances a patient’s priorities with existing medical literature and maximizes the likelihood of successful therapy.
8. Clinical workflow optimization
Hospitals perform a constant juggling act when managing patient inflows and outflows across various service departments, and always under the watchful eye of their budgeting department.
Patient flow optimization borrows some principles from supply chain management and logistics planning in that hospitals can build processes and systems for increasing the efficiency of an individual as he or she travels the entire length of the healthcare journey.
These steps include better routing of patients to the correct department for their respective conditions, forwarding them to any other necessary specialists, securing lab results, and potentially returning to the hospital should the need arise. AI can automate and manage much of the paperwork needed at every junction throughout the entire process.
Application of AI to medical patient flow may revolutionize this area of healthcare because algorithms can intelligently predict the ‘stickiest’ points in the process. Programs can forecast demand for emergency resources and prepare staff accordingly. Clinical care-focused AI applications in healthcare & the medical field can provide C-level executives with the ability to think ahead and adjust internal workflows to maximize efficiency.
An example might be to provide automated recommendations and reminders for patients with non-critical conditions to avoid visiting the ER when a regular appointment would suffice. Automating processes like this would avoid clogging up the arteries of the emergency room.
KenSci is a company that offers an AI-powered platform for optimizing hospital workflows. The patient flow management tool displays near-real-time visualizations of patient in-flows and outflows, giving administrators insight into total number of patients, length of stay, and overall capacity of the hospital.
9. Building intelligence into medical devices
The intersection of robotics, healthcare, and machine learning algorithms will catapult the medical device industry into the 21st century.
Automation of devices like ventilators and anesthesiology machines will be good targets because they rely on constant monitoring and adjustment of biofeedback signals including blood pressure, heart rate, and blood plasma levels of dissolved gases and pharmaceutical drugs.
This type of data is called time-series data. It is very difficult for humans to analyze effectively, but artificial intelligence tools can be trained to identify anomalies and crunch this type of data to effectively predict health problems.
Historically these types of AI applications in healthcare prove to be problematic from a regulatory perspective due to the inherent risk involved. However, as explainability of optimization algorithms advances and healthcare professionals can understand exactly why a machine makes the choices it does, these machines will grow to play a vital role within a hospital’s ecosystem.
10. The war against antibacterial super-resistance
Experts anticipate that antibacterial resistance will kill almost 10 million people annually by 2050. The problem is that microbes evolve resistance to medications faster than we can develop new drugs. The rise of MRSA (Methicillin-resistant Staphylococcus aureus) is becoming a widespread threat as indiscriminate use of antibiotics in homes, hospitals, and even agriculture continues to rise.
Artificial intelligence algorithms offer the possibility of identifying genetic weaknesses in drug-resistant strains of bacteria.
Research out of the Argonne National Laboratory uses an AI technique called Extreme Gradient Boosting (XGBoost) to explore the possibility of performing targeted treatments against these ‘superbugs’.
Initial results have been encouraging. The lab’s algorithm predicted with 95% accuracy which drugs and dose schedules would be most effective to treat bacterial strains with genomes the program has never encountered before.
To read more about AI applications in healthcare and the medical field, download this Health IT pdf.
Dynam.AI is ready to apply artificial intelligence to solve your healthcare problems
Dynam.AI offers end-to-end AI solutions for healthcare companies looking to incorporate the power of AI in their organizations. Our multi-disciplinary team of physicists, AI scientists, neuroscientists, and algorithm experts are fully equipped to help your team develop a roadmap for solving complex healthcare problems using machine learning.
Visit www.dynam.ai today to discuss the value that AI can unlock in your organization.
Subscribe to stay informed about AI applications in healthcare and business!
Dr. Michael Zeller
Dr. Michael Zeller has over 15 years of experience leading artificial intelligence and machine learning organizations through business expansion and technical success. Before joining Dynam.AI, Dr. Zeller led innovation in artificial intelligence for global software leader Software AG, where his vision was to help organizations deepen and accelerate insights from big data through the power of machine learning. Previously, he was CEO and co-founder of Zementis, a leading provider of software solutions for predictive analytics acquired by Software AG. Dr. Zeller is a member of the Executive Committee of ACM SIGKDD, the premier international organization for data science and also serves on the Board of Directors of Tech San Diego. He is an advisory board member at Analytics Ventures, Dynam.AI’s founding venture studio.