Understanding babies with AI: Ubenwa

Key points to remember:

  • Founded in 2017, Ubenwa is a startup that uses artificial intelligence to analyze babies’ cries and detect diseases and conditions such as birth asphyxia.
  • The startup recently closed a $2.5 million pre-seed and plans to use it for clinical validations and build an app that’s easy enough for parents to use.

Charles Onu spent five years studying Electrical Engineering at the Federal University of Technology, Owerri, Nigeria. But after moving to Canada in 2015, he made a foray into healthcare and technology and co-founded Ubenwa, a startup that detects early asphyxia in babies from the way they cry.

Onu was a researcher in machine learning at Mila – Quebec Institute of Artificial Intelligence and McGill University, Montreal, Canada, where he obtained his PhD in Computer Science – Machine Learning.

He founded Ubenwa alongside two other brilliant minds: Samantha Latremouille and Innocent Udeogu. Latremouille, a graduate of McGill University, studied translational biomedical engineering and holds a doctorate in experimental medicine.

Ubenwa Co-founders: LR: Samantha Latremouille, Charles Onu, Innocent Udeogu. Image source: provided.

Latremouille is the clinical manager at Ubenwa, while Udeogu is the software and engineering manager with over ten years of experience.

While the three co-founders are exceptionally good at their fields and undoubtedly qualified to develop technology to advance medicine, detecting disease through a baby’s cry still seems far-fetched. However, Onu spoke to Techpoint Africa to explain how Ubenwa works and the potential to save millions of babies every year.

Medicine man slash tech bro

Charles Onu, co-founder, Ubenwa
Charles Un

Onu’s specialization required him to work closely with doctors to think of new ways to use human physiological signals – heart rate, blood pressure and speech.

The development of new technologies that could use these signals for diagnosis led to the discovery of Ubenwa. Onu said the society was created specifically for babies because they cannot communicate like adults do.

“If I had a headache, I would say I have a headache. If a baby’s head hurt him, he would cry. If he had a stomach ache, he would cry too. The question was: ‘how can we improve communication with babies?’

According to Frontiers in Pediatrics, 2.5 million newborn deaths occur each year. Onu thinks this can be avoided if babies communicate as easily as adults. He added that failure to detect health problems early in babies leads to death or lifelong disabilities.

While Ubenwa wants to build technology that can detect all sorts of health issues in children at birth, it intends to start by perfecting one – birth asphyxia (BA).

BA is the inability of an infant to establish adequate breathing at birth, leading to decreased blood flow and oxygen levels in the body. In medical terms, it impairs gas exchange due to the interruption of placental blood flow.

BA represents the death of 1 million babies in the first month after birth.

The link between a baby’s crying and birth asphyxia

According to Onu, a baby’s cry is not voluntary like an adult’s speech; if there is something wrong with the body, the central nervous system triggers a crying response.

He discovered this while working with health-focused NGOs in Nigeria, such as Enactus.

“It was then that I first understood the seriousness of the infant mortality problem we have and began to examine it myself, adapting some previous work that I done on research on infant crying,” Onu said.

Interestingly, Onu’s drive to provide technological solutions to detect health problems early and accurately in infants began when his cousin suffered from birth asphyxia. Although his cousin survived, he developed a hearing loss.

With the help of doctors in clinical research, the team confirmed that the central nervous system coordinates babies’ cries. A baby’s cry used to tell what was wrong with the central nervous system became a logical possibility.

The next step was to collect data on the cries to develop an algorithm that would analyze the cries. When Onu and the team realized they could accurately detect – with a margin of error – birth asphyxia using a baby’s cry, Ubenwa, Igbo for a baby’s cry, was born in 2017.

This innovation meant they could detect health problems in babies before physical symptoms appear, in which case it is already too late.

Ubenwa wants to allow parents and even health professionals to detect the first signs of pathology in babies with everyday devices such as smartphones.

But it sounds like science fiction

Although Onu and the team found it easy to explain to medical professionals what Ubenwa is and how it works, convincing parents was a whole other story.

From his explanation, the problem seemed to stem from the state of technology. For most people, the detection of health problems should be more rigorous, such as analyzes or blood tests.

Simplifying the process to just listening to a baby cry felt more like magic. As a science fiction writer, Arthur. C. Clarke said, “Any sufficiently advanced technology is indistinguishable from magic.”

However, Onu understands why people might have doubts about the solution. He admitted that when he started working on the possibility of detecting birth asphyxia with a baby’s cries, he didn’t have high expectations of it.

He believes the proof will be to show people that it works. Ubenwa is currently collaborating with doctors in Nigeria, Canada and Brazil to acquire more data for clinical validation studies.

“We want to test on a large and diverse number of patients to prove that the technology works.”

How does Ubenwa work?

Pipeline for the identification of vocal biomarkers. Source: Karger

Using the sound of a baby’s cry to diagnose disease is just as fundamental as a malaria test. The main difference, however, is that while one looks for abnormalities in the blood, the other detects sound.

As Onu said, “we don’t look at biomarkers in the blood, but acoustic biomarkers.”

Research into the use of vocal or acoustic biomarkers has been ongoing for some time. This 2012 article titled Vocal acoustic biomarkers of depression severity and treatment response discussed how vocal acoustic properties in speech could serve as biomarkers of depression.

There is also research on the use of acoustic biomarkers to diagnose other diseases and disorders like Alzheimer’s disease, multiple sclerosis, and COVID-19.

When it comes to using sound to diagnose these diseases, Onu said the first step is to get the biomarkers and create a system that can calculate them.

This simply means collecting sound through a smartphone, laptop, or other device with a microphone, filtering out unwanted sounds, and processing it. Biomarkers are extracted via AI or an algorithm trained to look for pathologies in the baby’s cry.

The results, after biomarker processing, cannot be understood by just anyone. He must undergo another automated process that translates them, revealing the odds of having the neurological condition that was tested.

Even after the sound tests, a doctor must make the final decision as to whether or not the patient has a neurological disease.

Bringing Ubenwa to the public

Ubenwa recently closed a $2.5 million pre-seed round in June 2022, and Onu said a large portion of the fund would be spent on clinical validations to bring Ubenwa up to clinical standards.

The fund will also be used to create a mobile app for parents so they can know what their baby’s cry means.

Ubenwa is still in its experimental phase because the technology still needs a lot of data and clinical validations. Onu is confident, however, that the technology will find its way into more homes around the world and save infant lives.

About Hector Hedgepeth

Check Also

Folbigg’s life and advances in genetic research

New South Wales will conduct a second inquiry into Kathleen Folbigg’s convictions. At the center …