Microsoft is working with healthcare organisations around the world to help create pathways to more accurately diagnose diseases using artificial intelligence.
If you grew up on science fiction books and movies, you may be surprised to know that artificial intelligence is being used in industries around the world right now. In fact, it is transforming people’s lives in powerful and profound ways today. In manufacturing warehouses and factories, AI improves workplace safety by scanning products to find potential risks of injury to workers. In the United States, researchers are discovering ways that AI can assist public health organisations around the world to stop the spread of deadly diseases such as Ebola, Chikungunya, and Zika. AI can prevent them from flourishing at all by detecting the spread of pathogens in the environment to stop transmission to people before outbreaks begin.
This is the true future of AI — using new technology to build a safer and healthier world for everyone. Now that AI has given computers the capability to identify images and words, search for patterns in complex systems, and learn and reason similar to what people do, AI can enable devices to behave more responsively and naturally. This ability of AI is transforming how we as humans understand the world around us, and enhancing our human abilities and talents in a range of ways to enable us to start to find answers to humanity’s most urgent challenges.
Human health is an area in which AI can really be productive. AI makes it possible to envision a world in where we discover new pathways to address our most urgent challenges in healthcare, including chronic illnesses, cancer and heart disease. The good news is that scientists are already innovating possibilities for these pressing issues. From preventative care to personalised medicine and detection, AI is opening up limitless opportunities to reduce costs and improve outcomes for the healthcare field.
Microsoft is working with Apollo Hospitals, one of India’s largest private healthcare organisations, to use AI in order to improve the discovery of cardiac diseases that cause more than three million heart attacks in India every year. Before AI, it has been difficult for physicians to identify which patients are at risk for coronary disease because previous prediction models were based on research conducted in North American and Europe and didn’t apply well to people in India. In one example, high LDL cholesterol, a major cause of heart attacks in western countries, isn’t as common in India.
Now, researchers can combine deep expertise and rich data that Apollo Hospitals offer with Microsoft’s powerful AI and cloud capabilities to create a scoring system to discover patients in India who are at a high risk for suffering a heart attack.
To initiate this effort, a team of data scientists and Apollo clinicians began by reviewing over 400,000 patient records from all of its hospitals throughout the country and discovered that nearly 60,000 patients had experienced a heart attack after a visit for a health checkup. The puzzle was to identify the risk factors in Apollo Hospital’s data the earlier searches had overlooked. To learn the answer, the team uploaded all the data they had assembled to the cloud using Microsoft Azure, and then used Microsoft Azure Machine Learning services to look for hidden patterns.
Starting with 100 possible risk factors and 20 data points and using the computing power of the cloud, the team trained machine learning algorithms to search for the statistical significance of each risk factor in the occurrence of future heart attacks. This framework, allowed them to create a model which identified twenty-one risk factors prevalent in the Indian populations. Dr. K. Shiv Kumar, Chief of Cardiology at Apollo Hospitals, reported that the new model is 2X as accurate at predicting the chance of future heart disease as any previous models they had use. Not only is AI transforming how doctors conduct preventative health checks, but the same team is creating an AI-generated app to allow any person to find their heart-risk score without ever visiting a physician in their office.
Ray Zhang, CEO of the startup business Airdoc, assembled a team of engineers to create an AI-based tool for diagnosing signs of chronic diseases including hypertension, diabetes, age-related macular degeneration, arteriosclerosis by taking a high-resolution photograph of the back of the eye.
This new device takes advantage of the fact that examining the human retina is one effective method of assessing the health of the eye and to search for evidence of other illnesses. To develop the app, the Airdoc team used numerous retinal scans in order to create an algorithm using Microsoft Azure’s machine learning abilities which is trained to search for tiny abnormalities including spots, specks, deformed blood vessels, all of which can be warning signs for a range of health issues.
The Airdoc device is similar to the scanner used by optometrists for regular eye exams. To use it, a patient sits on a chair, places their chin on the padded brace, and looks into a special eyepiece. The algorithm automatically adjusts the angle of the device until a green cross comes into focus and then captures a high-resolution image which is instantly uploaded to the cloud. Once uploaded, the algorithm takes less than a second to conduct a detailed analysis which rates any susceptibility to numerous diseases as high, medium or low. Results are sent to the patient’s smartphone with a recommendation to seek medical help if they experience any potential problems.
At this point, the Airdoc system can identify signs of over 30 illnesses. Eventually, the device will be able to identify two hundred. Airdoc plans to make the device available in over hundreds of optical retails stores throughout China within the next few years. Using the device, physicians will significantly reduce the time spend assessing and reviewing scans, allowing them to focus on identifying patients that need treatment for chronic diseases in China and globally.
Microsoft is also working with Toronto’s Princess Margaret Cancer Centre at University Health Network (UHN) to redefine cancer treatment through a new approach titled “single cell sequencing” which enables physicians to analyse the genetic makeup of every cell within a cancerous tumour, and select the drugs that are optimised to kill the largest amount of cancer cells. Usually, physicians try one drug at a time to discover the most effective group of medicines for each patient. By using the power of Microsoft Azure Machine Learning and the cloud, single cell sequencing is giving doctors the ability to predict how each cell will respond to each drug available for the treatment of cancer, then develop a personalised therapy based on the specific genetic characteristics of each tumour.
Microsoft Azure provides a common platform to share medical data analytic tools with doctors and scientists across the entire country. Researchers at Princess Margaret Cancer Centre are envisioning a time in the near future when this type of genomic analysis is available for everyone in Canada.
These examples are just the start of the possibilities available with the collaboration of AI research and innovation that Microsoft and partners are involved in currently, and not just in healthcare.