Nature India | Indigenus

A predictive lifeline

“As a new doctor I always felt guilty when a baby died just after my watch, this takes away so much of that feeling, I can watch them from wherever.”

Runner-up, Nature India Essay Competition 2020

Abhilash Gangadharan

A neonatal ICU.

A. Gangadharan

I spun around at the noise. A baby in the critical section of the neonatal ICU (NICU) had woken from sleep, and began to heave and regurgitate into the ventilator mouthpiece. As I watched, it started gasping for air. To confirm my worst fears, the monitor began to beep as the baby’s oxygen saturation started falling.

I was an intern with a research group applying predictive analytics on real-time physiological data from neonatal infants. We were installing data acquisition modules that aggregated data from patient monitors and ventilators and sent them to the cloud for real-time analysis and predictions. We had already implemented a way for doctors to monitor the NICU from anywhere, and check on the status of each baby by logging into a website accessible only to them. The baby’s status, updated to the last minute, was at their fingertips even while they were away from the hospital.

There weren’t any doctors around now though. It was  2 a.m., and I was watching a baby die. I rushed out of the room and saw a nurse in the main NICU standing across the hall. I gesticulated wildly until she noticed me, she came over and I motioned to the baby. She looked at the clearly distressed baby and said “this isn’t my shift”. She went back to the nurse’s room and called out for someone else to deal with the problem. The baby’s struggle to breathe began to subside and with horror, I thought it would pass in front of my eyes. All I could do was look on helplessly as it choked on vomit.

Out of nowhere, a doctor came in. I pointed and croaked “baby…choking”. The doctor looked and yelled for two nurses and equipment. Over the next 10 minutes they suctioned away the vomit from its airways and replaced the contaminated ventilator paraphernalia and the baby was breathing again, peacefully.

Abhilash Gangadharan

Drained, I made my way out of the NICU into the hospital corridor. As I stripped my protective covers, apron and mask, the doctor came out and said “Thanks for the app”. It dawned on me then that his arrival at the critical moment was not just good fortune. He continued, “I was just about to leave, and gave the babies monitored by the app a final look, and saw this baby had turned red”, he was referring to the software indication of abnormal vital signs. He continued, “This is really going to change the way things are done around here. As a new doctor I always felt guilty when a baby died just after my watch, this takes away so much of that feeling, I can watch them from wherever. Can you put in an alarm system for when a baby goes critical as well?”

That was the moment of realisation for me. Initially we had met quite a bit of resistance from doctors who regarded this as an intrusion of analytics which made them feel as if their work was being audited. But as they got used to the system and saw how all the information for every patient was available at their fingertips, they grudgingly agreed it was convenient. But none of the doctors wanted any data-driven insights, medical decision-making or suggestions from our platform, clearly telling what they considered their prerogative. This was the first time a doctor had clearly indicated approval.

With an increasing number of premature births, more infants are required to spend time in the NICU. These are often under-staffed, with over-worked nurses and doctors. Expecting meticulous care and attention to detail for every neonate is hoping for the impossible. Medical complications are more common in neonates, and keeping track of a baby’s real-time vital signs is intensive. Avoiding errors in medication is possible by screening out contra-indicated drugs. By using the aggregated history of a patient’s stay in the NICU, it is possible to implement dynamic calculation of drug doses using daily birth-weights. This reduces the mental load of having to keep track of every patient and gives the doctor space and time to think of the bigger picture. This also helps them study and keep abreast of the latest in neonatal research, allowing them to improve their skills and processes of neonatal care.

The complementing of medical infrastructure with artificial intelligence-based systems improves clinical outcomes. The increasing number of premature birth cases is correlated with the rising number of IVF procedures due to fertility problems. More infants need to spend time in the NICU until they can safely be sent home. Improving the health outcomes of neonatal infants assumes manifold importance in such a scenario. Monitoring them constantly using automated algorithms and generating timely alerts is very useful for doctors. In case of any complications during their stay in the NICU, such a system ensures hypoxia is avoided. Such medical lapses can cause life-long problems for neonates. Predictive real-time analytics on incoming data can predict sepsis even before it is apparent to doctors using changes in heart-rate-variability, making timely live-saving medical interventions possible.

Ultimately, science that works towards improving the health of the nation’s youth would be the most impactful and help in improving society.

[Abhilash Gangadharan  is a PhD scholar at the Institute of Genomics and Integrative Biology, New Delhi.]

Suggested reading:

Announcing winners of NI Essay Competition 2020

Memories of paati

A grain of truth

Comments

There are currently no comments.