5 Ways Machine learning is Redefining Healthcare It has led to exciting new developments that could redefine cancer diagnosis and treatment in the years to come

Machine learning (ML) is an application of artificial intelligence (AI) wherein the system looks at observations or data, such as examples, direct experience, or instruction, figures out patterns in data and predicts events in the future based on the examples that we provide. Machine learning is seeing more and more use across industries for various reasons: vast amounts of data are being captured and made available digitally; processing of large amounts of data has become cost-effective due to the increased computing power now available at affordable prices; and various open source frameworks, toolkits and libraries are available that can be used to build and execute ML applications.

Specifically in healthcare, ML has led to exciting new developments that could redefine cancer diagnosis and treatment in the years to come. ML can increase access to treatment in developing countries which don't have enough specialist doctors that can treat certain diseases, it can improve the sensitivity of detection, add more value in treatment decisions, and it can help personalize treatment so that each patient gets the treatment that's best for them. In many cases they can even add to workflow efficiency in hospitals. The possibilities are endless.

Identifying Disease And Diagnosis

With growing populations and increased life expectancy, health systems are quickly becoming overburdened, under-resourced and not equipped for the challenges they face. Scientists have been working on ML models that predict disease susceptibility or aid in early diagnosis of diseases and illnesses. UK-based technology start-up Feebris is using artificial intelligence algorithms for the precise detection of complex respiratory conditions in the field. It connects to existing medical sensors and can be used by non-doctor users to identify respiratory issues early, avoiding complications and hospitalizations. In what could be an absolute game-changer, MIT's Computer Science and Artificial Intelligence Lab has developed a new deep learning-based prediction model that can forecast the development of breast cancer up to five years in advance. Their model was trained on mammograms and patient follow-up data to identify patterns that would not be obvious to or even observable by human clinicians. The results have so far shown to be far more precise, especially at predictive, pre-diagnosis discovery.

Medical Imaging Diagnosis

IBM researchers estimate that medical images are the largest data source in the healthcare industry. ML algorithms can process massive amounts of medical images at rapid speeds. And they can be trained to be extremely precise in identifying miniscule details in CT scans and MRIs. Companies such as Enlitic, Zebra Medical Vision and Sophia Genetics have developed ML algorithm-based analysis of all types of medical imaging reports and can diagnose malignancies or abnormalities with a higher accuracy rate than healthcare professionals. LYNA (LYmph Node Assistant) by Google detects spread of breast cancer metastasis early and can reduce the burden on pathologists as well. A deep learning convolutional neural network or CNN—developed by a team from Germany, France and the US—can diagnose skin cancer more accurately than dermatologists. In a recently reported study, the software was able to accurately detect cancer in 95% of images of cancerous moles and benign spots, whereas a team of 58 dermatologists was accurate 87% of the time.

The move from lab to actual practice has happened already for some AI-based solutions such as the FDA-approved imaging tool called IDx-DR for diagnosing diabetic eye disease.

Robotic Surgery

Robotics is changing the way surgery is performed today. The da Vinci robot is designed to facilitate complex surgery using a minimally invasive approach, reducing the length of surgeries and subsequently hospital stays. Various other robotic tools such as Stereotaxis in cardiac catheterization, Medtronic/Mazor in spine and neurology, Accuray in cancerous tumor irradiation, Stryker's Mako in orthopedic hip and knee replacement are improving surgical outcomes for thousands of patients. Even dental implants and hair transplants are being performed by surgical robots today.

AI and ML-based techniques will enhance the precision of surgical tools by incorporating real-time data, feedback from previous successful surgeries and data from electronic medical records during the surgery itself. This can help reduce human error and help general surgeons to perform complex surgeries in resource-limited settings lacking specialists.

Personalized Medicine

By applying AI and ML to multiple data sources—genetic data, electronic health records, sensor/wearables data, environmental and lifestyle data—researchers are taking first steps toward developing personalized treatments for diseases from cancer to depression. IBM Watson Oncology is making great strides in cancer treatment by leveraging patient medical history to help generate multiple treatment options. Similarly, a test named "CanAssist Breast' uses ML to identify a novel combination of biomarkers which play key role in recurrence of breast cancer. The test predicts the risk of recurrence for every patient. This helps personalize treatment by allowing patients with a low risk of cancer recurrence to receive less aggressive treatment.

Drug Development

ML can be applied at all stages of new drug discovery including designing the chemical/protein structure of drugs, target validation, investigating drug safety and managing clinical trials. The hope is that use of ML in drug discovery will not only help significantly reduce the cost of introducing new drugs to the market, but also make the drug discovery process faster (currently 10-15 years including clinical trials) and more cost-effective (currently costs almost $1 billion per new drug). AI company Atomwise's platform AtomNet uses deep learning software to sift through millions of possible molecules in a day or two, which would normally take months via traditional methods. The software then analyzes simulations that show how the potential medicine will behave in the human body. It has been able to identify possible medicines for multiple sclerosis and the deadly Ebola virus. Deepmind, the AI arm of Google's parent Alphabet Inc, is also making huge progress in this field.

Thus, it can be seen that AI indeed has tremendous potential and all stakeholders like the promising algorithms, accurate clinical and relevant in vivo data, clinicians, institutions have to align themselves to reap meaningful benefits from it.

One must remember that excellent technical innovations in AI can not fix social/political problems. Also the data input to AI must be in high volume and of clinically high quality/relevance. Fundamentally flawed data cannot substitute for high volume. Currently most of the AI applications are using the paradigm of "deductive reasoning' and we need to move from towards "inductive reasoning'.

We have travelled fair amount in the AI path to excellence but one must be cautious going further to embrace the brilliant promise it holds. What we need next is to move from theoretic benefit and evangelical sales to established use cases and robust, clinically-relevant data.