Tech companies have already expressed interest in the programme which has been tested cotside in a clinical trial involving more than 500 babies across eight European countries.
Prof Geraldine Boylan, principal investigator in the ANSeR study— Algorithm for Neonatal Seizure Recognition — said they have “trained the algorithm over many years to detect seizures”.
The success of the algorithm is proof of the usefulness of artificial intelligence (AI) in healthcare, she said, “not replacing jobs, but doing a job no-one else can do”.
“We hear a lot about AI taking over the world but there are certain areas in healthcare where we need this kind of help and this is one area where machines may do the job better,” she said.
The two-year trial, completed earlier this year, involved babies deemed at risk of seizure due to a difficult birth or who suffered hypoxic ischemic encephalopathy (HIE) — a type of brain injury that occurs when an infant’s brain does not receive enough oxygen and blood.
“All the infants were full term but there were clinical concerns that the brain was at risk of injury and we wanted to monitor their brain patterns,” said Prof Boylan.
She said seizures are difficult to detect in babies “because they don’t often show any visible signs”.
“But we need to know when to treat and this AI is like having a tireless eye at the cotside, constantly monitoring the baby’s brainwaves. An alarm goes off if the baby is having a seizure.
“Up to now, the standard monitoring for newborns at risk of seizure has been an EEG, a test that monitors the brain’s electrical activity. We have developed an algorithm that has allowed constant analysis of the EEG.”
The trial showed that seizures can be detected when the algorithm is used in real-time at the cotside, providing expert help “so machine learning does help”, Prof Boylan said.
“EEGs can be hard to interpret so having this expertise cotside will help clinicians pick up seizures as they are happening.”
The algorithm, which has been patented, was developed by a multidisciplinary team of doctors, scientists, engineers and computer programmers at the Infant centre in University College Cork. The trial using the algorithm, the first of its kind, involved the collection of thousands of hours of data.
Prof Boylan said developments in artificial intelligence, offer “limitless opportunities to support our work in the area of neonatal research, monitoring and neuroprotection for babies”.
The preliminary ANSeR findings will be presented tomorrow at the Brain Monitoring and Neuroprotection in the Newborn conference, underway in Killarney.
Prof Boylan is director of the INFANT Centre, professor of neonatal physiology at UCC, and conference host and co-chair.