Machine Learning and AI Technologies in the Healthcare Industry
One of the most important and fastest growing fields of our expertise is medical application of machine learning/deep learning and AI.
ML and AI technologies have been recently penetrating all spheres of healthcare services, from improving healthcare management to new drug discovery. Even though it is unlikely that computers will completely replace doctors and nurses, modern technologies are already transforming the healthcare industry as we knew it.
Only in 2016, over 100 companies were offering machine learning algorithms and predictive analytics to improve the effectiveness of the drug discovery process, provide assistance to patients, and diagnose ailments by processing medical images.
Here we offer some highlights of the machine learning applications that have emerged in the recent years:
1. Disease Identification/Diagnosis
A present-day human doctor cannot recall all the information necessary to make a prompt but an accurate diagnosis. With the modern abundance of textbooks, research papers and case studies, no doctor can master every aspect of medical care and recall every detail of the similar cases. Yet, it is possible to feed an AI-based system with relevant data and let the computer sift through the extensive database instead of relying on the much more limited human knowledge.
Disease identification was brought therefore at the forefront of ML research in medicine. Key players on the market were among the first to join the quest for the precise diagnosis, particularly in much-needed areas like oncology. To name only some of them, Boston-based biopharma company Berg applies AI to research and develop diagnostics and therapeutic treatments in multiple areas, including dosage trials for intravenous tumor treatment.
Another important example is Google’s DeepMind Health with its multiple UK-based partnerships, including with Moorfields Eye Hospital in London, in which they’re developing technology to address macular degeneration in aging eyes. The joint effort aims to find the early symptoms of visual problems caused by diabetes and age-related sight degeneration — the most important causes of sight loss in the UK. AI technologies will analyze more than one million eye scans and find out the first signs of visual degeneration which may be missed by most experienced doctors.
2. Personalized Treatment
Personalized medicine, which can provide for a more effective treatment based on the individual health data paired with predictive analytics, is also a hot research area. The dominant research method in this domain is so far supervised learning, which allows physicians to select from more limited sets of diagnoses and to estimate the given patient’s risk based on the similarity of symptoms and genetic information.
To give a single example, the U.K.-based health service Babylon raised $25 million in 2016 to develop an AI-driven app for healthcare services. It currently provides video consultations with doctors, while future versions will act as an AI-doctor: the application will receive the symptoms and check them against its database. Yet, in addition to the database, Babylon will consider the individualized history and circumstances of the patient, such as the family health history, medical records, daily habits, heart rate, cholesterol levels, allergies, and more. Some data will be obtained by monitoring the readings of the wearables.
3. Smart Electronic Health Records
Closely connected to personalized medical treatment is the area of health-related documentation. Document classification using support vector machines, as well as optical character recognition, i.e. transforming handwriting into digitized characters, are both essential ML-based technologies in the field of collection and digitization of electronic health information. The domain includes such big players as MATLAB with its ML handwriting recognition technologies and Google offering Cloud Vision API for optical character recognition.
The MIT Clinical Machine Learning Group has announced the development of state-of-the-art intelligent electronic health records withbuilt-in ML/AI which will simultaneously help with diagnostics, clinical decisions, and personalized treatment suggestions. MIT highlights the need for roburst ML algorithms which could “ learn from little labeled training data, understand natural language, and generalize well across medical settings and institutions.”
4. Behavioral Modification
The next decade will witness a rise of micro biosensors and devices, as well as mobile apps with more sophisticated health measurement tools and remote monitoring capabilities which will provide even more data to facilitate R&D and enhance treatment efficacy. This type of data will provide important insights to the individual in terms of their health optimization, but also will help reduce overall healthcare costs, if more patients adhere to following prescribed medicine or treatment plans.
Besides, behavioral modification is the core of prevention and there are plenty of start-ups popping up in the areas related to ailment identification, prevention, and treatment. To speak of some, Somatix, a data-analytics B2B2C software platform company offering an ML-based application which recognizes “hand-to-mouth gestures in order to help people better understand their behavior and make life-affirming changes”, specifically in smoking cessation. The second example, SkinVision, positions itself as a “skin cancer risk app” makes its claim as “the first and only CE certified online assessment.”
5. Drug Discovery/Manufacturing
Machine learning is a powerful tool in preliminary (early-stage) drug discovery which may be used in a range of activities, from initial screening of drug compounds to success rate prediction based on biological factors and to R&D discovery technologies like next-generation sequencing.
Startups are using machine learning algorithms to reduce drug discovery times, yet, the key players in this domain remain the MIT Clinical Machine Learning Group, who conduct precision medicine research focused on the development of algorithms to better understand disease processes for effective treatment of diseases like Type 2 diabetes and Microsoft’s Project Hanover who in cooperation with the Knight Cancer Institute is developing and AI technology for cancer precision treatment.
6. Clinical Trial Research
Machine learning has several potential applications which may help direct clinical trial research. For example, applying advanced predictive analytics in identifying candidates for clinical trials could provide a much wider range of data, including social media and doctor visits, as well as genetic information on the target populations. Better sampling techniques would lead to smaller, quicker, and less expensive trials.
ML can also be used for remote monitoring and real-time data access; for example, constant monitoring of biological signals for any sign of harm to participants or their eventual death. Some other possibilities for ML applications include finding the best sample sizes; addressing and adapting to differences in sites for patient recruitment; and using electronic medical records to reduce data errors.
7. Epidemic Outbreak Prediction
Another effective application of ML and AI technologies is monitoring and predicting epidemic outbreaks around the world. The data sources include data collected from satellites, historical information on the web, real-time social media updates, and more. For instance, support vector machines and artificial neural networks help predict malaria outbreaks on the basis of temperature, average monthly rainfall, total number of positive cases, as well as other data points.
Predicting outbreak severity is particularly crucial in third-world countries, which often lack medical infrastructure, educational opportunities, expertise, and access to treatments.
As Harpreet Singh Buttar, analyst at Frost & Sullivan claims, “by 2025, AI systems could be involved in everything from population health management, to digital avatars capable of answering specific patient queries,” and we at SciForce are ready to accept the ML and AI challenge with our current projects.