Research firm, Gartner, expects the global AI economy to increase from about $1.2 trillion last year to about $3.9 Trillion by 2022 , while McKinsey sees it delivering global economic activity of around $13 trillion by 2030 . And of course, this transformation is fueled by the powerful Machine Learning (ML) tools and techniques such as Deep Reinforcement Learning (DRL), Generative Adversarial Networks (GAN), Gradient-boosted-tree models (GBM), etc.
In this section, we will cite some key examples of modern application of AI/ML techniques in healthcare settings. Some of these are specific to the problem being solved, others are more generic in nature.
Did you know, and more than half a million deaths are reported from cancer? Breast cancer is listed as the top contributor of these cases. close to two million new cancer cases are diagnosed every year in the U.S. Deep learning (DL) and other classification and image processing techniques are being used in the fight against breast cancer (and other forms of cancers too).
MRI and other advanced medical imaging systems are being equipped with ML algorithms and they are increasingly forming the first line of checks for cancer detection. Often the number and quality of radiologists fall short in the face of overwhelming digitized data coming out of these imaging systems and ML-based systems are a perfect choice for aiding their decision-making process.
An article below has a comprehensive overview about various efforts in this regard.
ML techniques are being used for other forms of cancer detection too. AI has been improving the accuracy of breast cancer detection. A recent article from Nature, for example, describes how ML is applied to perform low-level image analysis tasks such as prostate segmentation and fusion of different modalities (for example MRI, CT and ultrasonography).
ML tools are also adding significant value by with information such as cancer localization during augmenting the surgeon’s displayrobotic procedures and other image-guided interventions.
Use of AI/ML techniques in assisting radiologists will only expand exponentially in the future. Throughout the world, and especially in the US, trained radiologists are under enormous strain due to deluge of digital medical data. A Nature article estimates that average radiologist needs to interpret one image every 3–4 seconds in an 8-hour workday to meet workload demands.
Imaging data is plentiful these days and DL algorithms can be fed with ever-expanding dataset of medical images, to find patterns and interpret the results as a highly trained radiologist would. The algorithm can be trained to sort through regular and abnormal results, identifying suspicious spots on the skin, lesions, tumors, and brain bleeds, for example.
These kind of ML methods can be particularly helpful in identifying rare or difficult to diagnose diseases as they are built on large datasets containing images of these diseases and they are often more dependable than humans when it comes to detecting ‘ edge-cases ‘ (if trained properly).
For example, Microsoft’s Project InnerEye employs machine learning to differentiate between tumors and healthy anatomy using 3D radiological images that assist medical experts in radiotherapy and surgical planning. In their own words, it can help in the following ways:
extraction of targeted radiomics measurements for quantitative radiology,
efficient contouring for radiotherapy planning,
precise surgery planning and navigation.
An ever-increasing amount of medical data are being digitized at all public and private healthcare institutions (hospitals, doctors’ clinics, pathology centers, etc.). However, by their very nature, this kind of data is messy and unstructured. Unlike other types of business data, where traditional statistical methods can be used for quick insights, patient data is not particularly amenable to simple modeling and analytics tools.
We need highly dependable and robust platforms which can connect to a multitude of patient databases and analyze a mixture of radiology images, blood tests, EKGs, genomics, patient medical history, etc. Furthermore, they should be able to distill down their analyses and the hidden patterns that they discover, to human-intelligible forms so that doctors and other healthcare professionals can work on their output to prescribe affordable and responsible patient care regimes.
For example, Enlitic, a San Francisco based start-up , has a mission of mixing intelligence with empathy and leverage the power of AI for precisely generating such insights, as discussed above. In their own words, “ by pairing world-class radiologists with data scientists and engineers, we collect and analyze the world’s most comprehensive clinical data, pioneering medical software that enables doctors to diagnose sooner with renowned accuracy .”
The cost and difficulty of receiving proper health care has been a subject of long and bitter debate in the US. Americans, on average, visits doctors only slightly above four times a year, which is much less compared to that in other developed nations (13 times for Japan, for example).
What are the pressing issues for avoiding healthcare systems? Long queue, fear of getting unreasonable bills, frustrating appointment process, not getting paired up with the right professional.
The key here is to develop and employ suitable AI-assisted platforms whose primary mission is to enhance the experience of healthcare services for the largest section of common people. As mentioned above, traditional businesses in many sectors, have started large-scale deployment of such tools for improving their operational efficiency. But most often, the overarching goal of such platforms is to maximize profit for the businesses. The distinguishing feature for adopting similar powerful AI tools for healthcare organizations will be to carefully mix the aspects of empathy with the goal of profit generation.
What are the possibilities of AI and ML being used for surgical robotic assistants? A recent article from RoboticsBusinessReview has explored the following topics for surgical robotic assistants :
Distributed data-processing and software-centric collaboration between a team of surgical robots
AI-generated virtual reality space (built based on thousands of historical images and 3D data streams) which can direct and guide a surgeon real-time
Data-driven insights and advice based on previous surgery histories (not only performed by machines but by humans)
Increasing possibility of fusion with tele-medicine and remote surgery for relatively simple procedures
U.S. pharma companies, on average, spend over 2.5 billion dollars per successful drug development and market launch. This stiff price tag, no doubt, impacts the overall cost of healthcare for the general population.
AI and ML techniques are increasingly being chosen by big names in the pharma industry to solve this hellishly difficult issue. AI and ML is very effective at hunting for new pharmaceutical opportunities in the industry in examples like:
Sanofi has signed a deal to use UK start-up Exscientia’s AI platform to hunt for metabolic-disease therapies,
Genentech is using an AI system from GNS Healthcare in Cambridge, Massachusetts, to help drive its search for cancer-related treatments,
Pfizer is using IBM Watson platform and its vast array of analytics and ML engines, to power its search for immuno-oncology drugs.
Apart from the long-haul chain of drug discovery, fundamental processes of early-stage candidate selection and even mechanism discovery are being thought to be particularly amenable to AI techniques. For instance, Researchers at the biotechnology company Berg are working on a model to identify previously unknown cancer mechanisms using tests on more than 1,000 cancerous and healthy human cell samples and tracking data from their lipid, metabolite, enzyme and protein profiles. The group uses its AI platform to generate and analyse immense amounts of biological and outcomes data from patients to highlight key differences between diseased and healthy cells.
The development of high-throughput, data-intensive biomedical research assays and application of large-scale statistical modeling to such datasets have already revealed a great heterogeneity in the pathophysiologic factors and processes that contribute to disease, elucidating the fact that there is a need to tailor, or personalize , medicinal regimens to the nuanced and often unique features possessed by individual patients.
According to the U.S. National Library of Medicine, precision medicine is “ an emerging approach for disease treatment and prevention that takes into account individual variability in genes, environment, and lifestyle for each person. “ And this is one of the biggest anticipated benefits from application of AI and ML in healthcare.
1. How to train a neural network to code by itself ?
2. Paper repro: “Learning to Learn by Gradient Descent by Gradient Descent”
3. Reinforcement Learning for Autonomous Vehicle Route Optimisation
4. Artificial Intelligence Conference
And, according to a recent Forbes article , the added advantage of such an AI-driven system will be the ability to not only predict and model outcomes, but also to predict future patients’ probability of having diseases given early screening or routine annual physical exam data. By being able to model why and in what environments diseases are more likely to occur, AI tools can prepare medical professionals to intervene (in a personalized manner) before a disease is showing symptoms.
In this article, we discussed a wide variety of exciting applications of powerful AI/ML techniques and platforms, built on those algorithms, in the space of healthcare. And we also discussed topics ranging from pathological assistant, to operations management, to personalized medicine, and drug discovery.
There are, of course, known challenges from data privacy and legal frameworks. Health and patient related data are some of the most closely guarded secret and it is extremely complex to figure out what can be used legally by private third party tool providers (in this case the owner of the AI and ML tools or platforms).
Therefore, a massive parallel effort to rationalize the legal and policy-making is needed to bring the full benefit of advancement in AI technologies into the healthcare space. As technologists and AI/ML enthusiasts, we can only hope for such a bright future where the power of these intelligent algorithms benefit millions of common people to improve their well-being.Don’t forget to give us your 👏 !
