When life gives you lemons, you can do whatever you want with them. Problems, on the other hand, require solutions that come from deep thinking and, sometimes, experience. For Canada-based Charles Onu, the problems he experienced in Nigeria formed the womb that would later produce Ubenwa, his innovative healthtech startup that leverages advanced technology to preserve the lives of newborns.
In most parts of the world, immediately after a child is born, it undergoes a test called APGAR scoring. This test is a non-invasive examination of the baby that helps to identify whether the child has a medical condition to be treated. At that stage of the baby’s life, though, almost every illness can fall under the life-threatening category, which is why early diagnoses are imperative.
Regrettably, APGAR scoring is a lagging indicator ( a method that diagnoses some medical conditions after they’ve progressed to an extent). It has been unable to detect life-extenuating conditions like birth asphyxia as early as is necessary. This delay results in the deaths of infants annually. Onu’s cousin was one of such newborns that suffered from birth asphyxia in Nigeria. Thankfully, he didn’t die, but he later developed a hearing condition as a result.
Years after that experience, Onu found himself working in the health department of Enactus, an NGO that helps to develop young student leaders. His work there exposed him to the rigours of neonatal care, and again, he saw firsthand how late detection of medical conditions costs infants their lives.
He later moved to Canada and spent seven years working closely with neonatal experts in ICUs, only to realise that the problems he saw in Nigeria were actually global. It is difficult to communicate with infants, and this inaccessibility puts their lives at risk. The world needed a way to understand babies’ health from their first cries, but someone had to build the solution for that. That someone, Onu didn’t realise, would eventually be him.
Founded in 2017, Ubenwa is a software-as-a-service company that provides the technologies—mobile app and API— that enable the medical interpretation of infants’ cries, using artificial intelligence and machine learning algorithms to extrapolate patterns and insights. The startup is a spinoff of five years of research at Mila, a world-renowned AI hub in Quebec, Canada.
Onu, on a call with TechCabal, asserted that at Ubenwa’s core is a determination to detect diseases in infants early enough to preserve their lives.
“We want to focus on this objective of early detection. If we can detect conditions early, then we can prevent long-term diseases. We can also prevent the high rate of infant mortality that exists today.”
“Our first set of products is an app and an API that allows parents and physicians to use Ubenwa via their smartphones or video devices like baby monitors. With these, they can connect with our technology to carry out early screening for neurological problems,” he said.
Ubenwa’s founding team comprises Onu, an audio machine learning scientist and AI specialist who doubles as founder and AI lead; Samantha Latremouille, an award-winning scientist in experimental medicine who works as the clinical development lead; and Innocent Udeogu, the software engineering lead with over 10 years of startup experience. The trio are building on work previously done in the field since the 1970s when researchers found that unusual cry patterns of infants were indicative of health issues.
“When we started Ubenwa, one of the things we did was to start to speak with one of those early clinicians. One of them, now an elderly guy, told us how they couldn’t save the cries because they didn’t have enough data storage systems. They would record the sound, extract the features, and lose the sample. But since research and experimentation are iterative processes, they couldn’t go back to the data to extract new features and analyse them.”
According to Onu, Ubenwa has been able to push through the limitations of the previous decades by leveraging the modern technology of this age, including AI, ML, and cloud systems. Now, the startup claims to have the largest and most diverse database of infant cry sounds that have been clinically annotated.
Another of Ubenwa’s victories was its successful pilot to detect neurological injury due to birth asphyxia. Ubenwa’s software showed a 40% improvement over APGAR scoring.
Responding to why the technology currently focuses on neurological injuries, Onu said:
“There is a whole swath of medical conditions that we’d have to zoom into different stages of the clinical process, but presently, we’re promoting our work on neurological injuries. This is because neurological diseases like birth asphyxia, which reduces oxygen supply to the brain, are responsible for a third of all newborn mortality globally. So it was clear to us that solving this problem was a way to make the biggest impact,” he added.
Ubenwa’s lofty ambition to power a standardised test for babies in the future has not been without struggles. Onu described the lack of pre-existing data as one of his team’s major challenges from the beginning. They were building a global solution, so they needed a widely diverse data representation to test their research. However, this data was difficult to lay hands on, as no hospital previously recorded cry samples for scientific research.
Ubenwa is currently headquartered in Canada with partners in Nigeria and Brazil. Hospitals in Ubenwa’s network include Montreal Children Hospital, Canada; Enugu State University Teaching Hospital, Nigeria; Lagos State University Teaching Hospital, Nigeria; Rivers State Teaching Hospital, Nigeria; Santa Casa de Misericordia, Brazil; and Lagos University Teaching Hospital, Nigeria.
The startup raised a $2.5 million pre-seed earlier in July 2022, which Onu says will be used to complete the clinical validation of its technology and to launch a private beta testing for real-life applications.
Charles Onu and his team are daring to bet on AI for babies. Maybe they’re making lemonades from lemons, after all.
Inside NVIDIA plans to develop Africa’s AI ecosystem