Nigerian clinical speech recognition startup Intron Health is aiming to bridge the gap in speech recognition for minority languages and accents, particularly in healthcare, and has just raised sizeable funding to help it do so.
Launched in 2020 by Tobi Olatunji and Olakunle Asekun, Intron Health developed Africa’s first clinical speech recognition platform, which boasts up to 92 per cent accuracy rate on medical terminology with heavy accents.
The platform helps doctors across Nigeria, Ghana, Kenya, South Africa and most recently Uganda complete documentation seven times faster, significantly accelerating the adoption of Electronic Health Records (EHR) and reducing the administrative burden.
Its speech recognition tool is trained on 3.5 million audio clips from 18,000 contributors, representing 29 countries and 288 accents. Intron provides automated real-time speech transcription services to doctors at busy hospitals helping to reduce documentation time, patient wait time, and accelerate the transition to digitised care.
Olatunji is an ex-AWS machine learning scientist who initially received medical training and practised as a medical doctor in Nigeria, but later pursued a master’s degree in medical informatics from the University of San Francisco and another in Computer Science at GeorgiaTech. He explained to Disrupt Africa how Intron Health came about.
“The company was formed with the vision of transforming healthcare delivery across Africa by leveraging artificial intelligence and machine learning technology upon recognising the unique challenges of documentation and administrative tasks faced by the healthcare sector on the African continent,” Olatunji said.
The significant gap in the market Olatunji and his team had spotted was that existing speech recognition tools struggled to accurately transcribe African accents and medical terms, leading to inefficiencies in healthcare settings.
“As hospitals transitioned from paper to computers, the cognition load and slow typing speed with keyboards became a significant bottleneck to adoption by already overworked clinicians,” he said.
The startup’s tool has been adopted in 30 hospitals across five markets, with positive outcomes, such as reducing turnaround time and patient wait times for radiology reports and general hospital administrative activities. It is now set for further growth, after securing a US$1.6 million pre-seed round of funding, led by Microtraction, with participation from the likes of Plug and Play Ventures, Jaza Rift Ventures, and Octopus Ventures.
“With this funding, Intron Health will deepen its research efforts, strengthen cloud-native and on-prem capabilities, and expand distribution. The company will also bolster its team by recruiting tech talent to support product development and market expansion, driving continued progress and breaking further technological barriers,” said Olatunji.
In addition, Intron Health has partnered with Google Research, the Bill & Melinda Gates Foundation, and Digital Square at PATH to evaluate the effectiveness of large language models – like ChatGPT – in African healthcare.
“It is the largest study on LLMs in global health, creating a novel dataset of 25,000 Africa-focused exam-style question-answer pairs covering 32 medical specialties from over 1000 clinical contributors from 15 countries.
The startup has plans to expand to new markets and explore new growth frontiers post-funding, and generates revenue by offering its speech-recognition tool to healthcare providers through subscriptions or licensing models.
