Avin Aggarwal, 34, a medical electronic engineer, and Ashim Roy, 62, a telcom engineer, had always wondered about the possibilities of diagnosing and treating diseases without doctors. Not because they did not believe in the competence of doctors, but they thought that in a country awfully short of doctors, intelligent devices could replace their intervention in many cases.
So, on a bright February afternoon in 2014, the two friends decided to combine their professional experience to develop a portable device that would allow physicians in remote areas to easily pick up cardiac abnormalities.
In India Electrocardiography (ECG) has been the standard method to detect any abnormality in heartbeat. Though there are many portable ECG devices available in the market, you still need a skilled cardiologist to read the graph.
But that might change soon!
After three years of research, the duo’s new artificial intelligence-based ECG device is ready. Its makers say its diagnosis of heart ailments is as accurate as that of a cardiologist.
Ashim Roy and Avin Aggarwal of Uber Diagnostics, which has made Cardiotrack, an AI-based ECG
"We realised that a portable ECG would not help unless there are enough trained doctors or technicians to interpret an ECG graph. Unfortunately in India where about 50 million people suffer from heart ailments, there are only around 8000 cardiologists," says Aggarwal. "So, we developed algorithms and fed them with data of over half-a-million cardiac patients from across the world," he says.
Their self-sufficient smart device uses this data to read the abnormality in the ECG of a patient.
To test its accuracy, they asked the doctors at Fortis hospital, Noida to use it in their emergency department for a month. "It was a small, handy device, which I thought could be really useful in ambulances and rural areas. It can read the ECG and the App attached to it can send the ECG report on your mobile. One can diagnose a heart attack with it and take necessary steps to prevent patient's life," says Dr Dina Shah, head, emergency medicine, Fortis Noida.
Cardiotrack device
In fact, there are many such smart AI- based devices that have the potential to save many lives. They work as a specialist doctor where there is none. Take for example Automatic External Defibrillator (AED), which have been installed at many airports and public places across the country. In case of a cardiac arrest, this machine can detect if the patient needs shock treatment and guides the person attending to the patient on how to give it. Similarly, automated CPR machine—another AI-based device can resuscitate a dying patient. Experts believe that artificial intelligence has the potential to revolutionise healthcare, especially in India where there is a severe shortage of skilled manpower in healthcare.
Dr Gautam Yadav, a pediatrician in Rewari, a small town in Haryana, has a neonatal unit at his hospital. He uses a smart AI-based platform, integrated Neonatal Intensive Care Unit (iNICU), to track the treatment of all the children admitted at his hospital.
Unlike a nurse, the device reads data from all the monitors attached to the newborn continuously. Besides, it keeps track of all the pathology reports, blood tests and medicines being given to the child. "It is a foolproof system that doesn't leave any room for mistake. It alerts if there is any fluctuation in the parameters," he says.
iNICU was developed by Harpreet Singh and his wife Ravneet Kaur of Oxyent Technologies with the objective of predicting the extent of infection and any other abnormality in neonates. The technology will help doctors prepare the course of treatment to deal with any complication because neonates don't give much time to their doctors. "In India we lose 39 babies out of 1000 admitted at NICUs. Predicting infection, I believe, will save many lives. The ultimate aim is to predict sepsis, the most fatal infection in newborns, a week in advance. That will give a lot of time to doctors to prevent deaths," says Harpreet.
But how do these artificial intelligence based devices work? Artificial Intelligence is basically complex computer software programs that replicate human neural intelligence. Computer scientists create an algorithm to predict a disease. Now, a lot of data is fed into this algorithm --blood reports, records of medicines, imaging scans, pathology reports, doctor's clinical diagnosis -- from patients who have suffered from or are suffering from a disease.
Now the system can identify the symptoms and physiological changes that occur before, during and after in a patient suffering from that disease. "The clinical diagnosis formed by this platform could be as or more accurate than that of a highly skilled clinician. And their accuracy improves with the amount of data they are being fed," says Sanjay Singh, CEO , QorQl, a Noida-based software company that makes AI- based technologies for hospitals and doctors.
"AI works on two data and algorithms. In India, we never gave importance to data generation before. Right now, we are working on collecting and aggregating data from different sources. Soon, we would have a lot of AI-based solutions," he says.
Agrees Dr Samir Brahmachari, a noted genome scientist, "Look at the population of India and the diverse genetic pool that we have in India. The data collected in India will be revolutionary."
Genomics, doctors feel, is another area, where AI can help a lot. The biggest challenge in molecular biology is to extract DNA from blood, or saliva, or clinical tumour of a cancer-affected tissue and then read the code of the disease in the DNA.
Dr Vijay Chandru at his lab at Strand Life Sciences
AI can help researchers find the code by going through the data worth kilobytes in a few minutes time. "To interpret a gene sequence, you need intelligence, context, understanding, and knowledge of what other scientists across the world think of that particular mutation in the gene. An AI-based solution is capable of all this," says noted computer scientist, Vijay Chandru of Strand Lifescience.
Strand Lifesciences has built a platform called 'Grammatica', which can not only read scientific literature, but can also extract the co-relation between a genetic pattern and the biology of the disease. Besides, it can tell how a person would respond to a particular treatment.
The company has also built a robust data mining technology that it has named after A K Ramanujan, renowned Hindu theologian and philosopher. The technology may help in early detection and treatment of some genetic diseases.
AI is all about finding relationships among large data sets to tease out predictions. It will change the way medical records are assessed and kept.
In a study conducted in 2016 by Frost & Sullivan, the market for AI in healthcare was estimated to reach 6.6 billion dollars by 2021. AI, health experts and computer scientists feel, will strengthen medical imaging diagnosis and care delivery. Some say that it has the potential to improve outcomes by 30 to 40 percent and may reduce the costs of treatment by 50%. “AI is the next big thing after genomics in medical care. If we can combine the two, the outcomes will be revolutionary. Accurate diagnosis and effective drug are the key to cure,” says Brahmachari.