Listen to this story
The healthcare industry has long been a hotbed for AI innovation, pushing the envelope on possibilities. The industry has long suffered a talent crunch, a problem that AI could help solve. However, using artificial intelligence and machine learning in this field comes with a host of challenges.
First, the medical data of users must be preserved at any cost. Medical data is one of the most sensitive forms of personally identifiable information and requires the highest security and confidentiality possible. Second, AI solutions cannot be allowed to make decisions on their own when it comes to the status of patients.
Notwithstanding these roadblocks, automation through AI can vastly improve the quality of life for medical professionals. Analytics India Magazine reached out to Aneesh Nair, CIO & co-founder of MyHealthcare, a speciality health tech company launched in 2017, to learn more about how AI is being deployed in the medical field.
AIM: Medical data is one of the most important areas to implement data privacy and protection. What measures does MyHealthcare take to protect users’ data?
Nair: As a cloud-native, and mobile-first healthcare application provider, we leverage new-age cloud modules to deliver the gold standard in security for our customers. To protect patient data, we use Data-at-Rest and Data-in-Motion Encryption, which is one of the most effective ways to protect it from unauthorised access. This makes it much more difficult for attackers to access or steal the data.
Our ecosystem has also implemented strict access control measures as another important step in protecting patient data. This is done using a variety of techniques, such as isolation, authentication and authorization, to ensure that only authorized users can access the data.
Conducting regular security audits is another important measure that we use to ensure the privacy and security of patient data. We also ensure that all software and security patches are installed promptly to protect against the latest threats. By making multiple copies of patient data and storing them in different locations, we ensure recovery of the data in case it gets lost or corrupted. We also provide regular training to our client’s employees on security and encourage them to report any suspicious activity.
AIM: What are some of the biggest optimisations you have seen deploying AI applications in your application?
Nair: MyHealthcare uses AI to help automate many routine tasks such as data entry and analysis. This can save time and resources, allowing healthcare providers to focus on more important tasks and deliver care more efficiently. AI helps us analyze a patient’s medical history and other data to create personalized treatment plans. This would help doctors tailor treatment protocols to the specific needs of each patient, leading to better outcomes and a more satisfying experience for the patient.
We can also analyze large amounts of data with the help of AI to identify trends and make predictions about future health outcomes. This can help our healthcare providers identify potential problems early and take steps to prevent them, improving overall health and reducing costs. Our ecosystem also uses AI to analyze medical images and other data to identify patterns and make more accurate diagnosis.
By implementing Speech to Text feature in our IPD & OPD EMR applications, we are reducing the doctors’ burden by providing them the speech option to enter all the patient details like chief complaints, history of present illness, review of systems, current medications and allergies. Similarly, to better serve the patients, we have a real time analytics engine implemented to improve operational efficiency across our platforms. Further to get the 360-degree view of patients, we present the patient data readily available for the doctors through a patient journey mapping.
The system mentions the initial symptoms, tests advised, diagnosis done, medications and follow up details, all of which are made available on a single screen for the doctors. Overall, the benefits of using AI in our applications are numerous and have a positive impact on patient care and outcomes.
AIM: Can you delve deeper into the specifics of how you have deployed AI in your enterprise ecosystem?
Nair: We have successfully deployed AI-assisted tuberculosis detection modules for the Radiology department, where a computer vision-based application trained on thousands of X-ray reports, works on patient x-rays and identifies any abnormalities in the lungs. The algorithm also highlights them through heat maps and provides the probability score for the doctors to investigate further.
Such systems identify the lung nodules that even an experienced radiologist might miss with his naked eye. Similarly, the system is able to accurately identify abnormalities in different locations and even behind the bones. This helps doctors make better decisions about the treatment, leading to improved patient outcomes.
In addition to TB, we have also developed an application to detect Parkinson’s disease, a neurodegenerative disease, which causes patients to suffer from movement disorders, making it difficult for them to manage their frequent hospital visits.
Our mobile app makes it possible for patients sitting in the comfort of their own homes to send their walking videos to the AI engine. Here, we apply human pose estimation techniques to extract the body coordinates and compute the variability using a trained deep learning model. This then provides the probability score for the doctors to decide on the next course of action.
AI is used extensively to analyze electronic health records to identify trends and make predictions about future health outcomes. This helps doctors get a longitudinal view of patient records and identify potential problems early and take pre-emptive steps to prevent them. We also use NLP algorithms in our applications to analyze patient notes and other text data to extract information and gain insights. Further, on the radiology side to automate the report generation, we have deployed our specially trained radiology speech engine.
This solution understands the complex radiology terminology dictated by the radiologist and accurately converts the speech into text, thus improving the efficiency of the radiologist. Currently, we are working on building a comprehensive clinical decision support system which provides references and suggestions to doctors for the given symptoms.
AIM: There are also many risks arising from using AI in healthcare. Have you faced any?
Nair: While using artificial intelligence in our applications brings many benefits, we are aware that there are also potential risks associated with unsupervised AI. The way we use AI (Augmented Intelligence) is to augment the capabilities of doctors and healthcare professionals and never use it as a tool for replacing doctors.
We always ensure there is a human in loop for all decision-making processes where AI is involved. Further by design, we ensure the ethics and fairness consideration are part of our machine learning model, so that we avoid any potential errors and biases towards the decision-making process. By carefully managing the potential risks and taking steps to mitigate them, we ensure that AI is deployed through our applications systems and is used safely and effectively.
