Finding relationships among large data sets to tease out predictions — or ‘big data’ analysis — is the hottest trend among e-commerce and biotechnology companies. New research seeks to extend it to neonatology and the fight against preterm baby mortality. Since January, Oxyent Medical — a company run by Harpreet Singh and his wife Ravneet Kaur— has been collecting data from about 300 children tracking their progress over their months in the neonatal intensive care units (NICUs) in Delhi hospitals. “We’ve made an app out of it and now parents can track the progress and the weight gain. Going ahead, it can also serve as a vaccination record,” says Mr. Singh. The hope is that over time, long-term medical records of a range of Indian preterm babies will be available and researchers can then begin to see if there are correlations between, say, the weight and heart rate at 25 weeks and if that in turn can reveal patterns and predict if a child is progressing to sepsis. That’s shorthand for an array of bacterial infections that do not affect infants who’re born at 32 weeks or after but almost always kill those born even three-four weeks shy.
Mr. Singh’s interest in this stems from personal tragedy. Life in 2010 was near-perfect for the then 31-year-old. Two advanced degrees in biomedical science from the University of Wisconsin and a publication in Science , the world’s top research journal. A software for improving brain scans that EGI, an American biomedical company, bought. Relocation thereafter with his wife to India, followed by the couple launching their own company. Soon they were in the family way, Ms. Kaur expecting twins. Then, they lost one of them. She had gone into preterm labour after 26 weeks of gestation, instead of the normal 34-38 weeks. Though one of the children survived, the doctor said that the deceased infant — barely bigger than the palm of an adult — had succumbed to sepsis.
Keeping a constant vigil Dr. Raghuram Mallaiah of Fortis La Femme Hospital in Delhi, who specialises in tending to preterm children, says that the challenge in treating such babies is in being able to identify the onset of infection early enough and begin appropriate antibiotic treatment. “The class of bacterial infections that we see in India are frequently different from that in the West,” says Dr. Mallaiah, who’s worked in England for a decade, “and that complicates treatment.” A general distinction between bacteria is whether they are ‘gram positive’ or ‘gram negative’ and medical literature finds the latter to be more resilient to antibiotic treatment. “Gram-negative infections are far more common in India and we barely get a 24-48-hour window to treat them,” he adds.
Moreover, in such fragile babies weighing barely 1.5 kilos, the dosage of the administered antibiotics matters greatly because too much could mean further weakening them. Medicine apart, all other vital nutrition for such babies is also calculated based on their weight and so it’s one of the parameters that’s obsessively monitored for preterm babies. “Many times there are too many babies and too few nurses and they have to keep scanning other vitals such as urine, heart rate, lung function on the several monitors… errors happen,” says Dr. Mallaiah. Mistakes then mean that that the attending doctors too don’t have the right numbers to suggest subtle changes in dosage or nutrition specific to each child.
To take a shot at minimising such errors Dr. Mallaiah and Mr. Singh, who’d spent nearly two months amidst the grim beeping machines that marked the progress of his twins, decided to combine their skills. As part of their ongoing project, Mr. Singh used his experience as a biomedical engineer to integrate the several machines monitoring each child such that the latest vitals — especially weight — were automatically updated and available to doctors.
The importance of scale Parameters such as the head circumference, length, weight, blood oxygen level from a small pool of infants at Fortis and the All India Institute of Medical Sciences are part of an initial tranche of data that has so far been collected, which will serve to ‘teach’ Oxyent’s systems to find correlations among these numbers to predict serious contingencies. “Every day, every baby generates about 1 gigabyte of data,” says Mr. Singh. Implementing such a data collection system is feasible even in government hospitals provided it can be integrated within a cloud-like enterprise.
The duo’s ongoing research has a global precedent. In 2011, researchers at the University of Virginia developed a system called HeRO (Heart Rate Observation) that measures subtle fluctuations in the heartbeat of infants and gives them a score from 1-5 to calculate the odds — the higher the number, the greater the odds — of them going into sepsis. There haven’t been many trials but reviews suggest that the system has had some success. For instance a study in Research And Reports In Neonatology records the case of an infant born at 25 weeks’ gestation. When she was four weeks old, doctors saw a slight increase in the infant’s temperature and heart rate, and initially attributed it to overheating by the incubator. But with her HeRO score increasing from 0.8 to 2.0 in six hours, blood and urine cultures were obtained and antibiotics administered. In 24 hours, the doctors found that she was infected by the Klebsiella pneumoniae bacterium and put her on a full course of antibiotics.
Big data, if it ultimately yields results for neonatology, can make a big difference in India. The World Health Organisation says that globally 1 in 10 live births are preterm, > with India registering the highest number of them at 3.5 million . Even accounting for India’s high population, it’s still thrice that of China — the world’s most populous country — which has 1.1 million such births. Nearly a tenth of those children die due to complications and again, among 199 countries studied by Save The Children foundation in 2012, India has the highest number of deaths. But it is early days still. “The system isn’t developed enough to be deployed in hospitals,” says Dr. Mallaiah, “but we still need data from Indian babies to even hope to use them.”