Intron Health Secures Support for Its Speech Recognition Solution Tailored to African Accents

The integration of voice recognition technology into the fabric of daily life is more widespread than ever, yet it significantly underserves speakers of less common languages, those with strong accents, or individuals with speech impediments, such as stuttering. These groups often find it challenging to engage with voice-activated systems designed for tasks like application control, transcription, or automation.

Tobi Olatunji, the innovator behind the startup Intron Health, is on a mission to close this divide. Boasting what he describes as Africa’s most extensive clinical speech database, Intron has refined its algorithms using 3.5 million audio samples (equaling 16,000 hours) contributed by over 18,000 individuals from 29 countries, encompassing 288 different accents. The vast majority of these contributors are professionals in the healthcare field, ensuring precision in the capture and pronunciation of medical terminology for the regions Intron serves.

Olatunji comments on the advantage this brings, particularly for those in countries like Ghana, Uganda, and South Africa where data input is rapidly increasing, saying, “Our significant training on African accents means users here are likely to experience superior baseline performance over other systems.” He expresses confidence in the technology’s deployment within these markets.

His journey into health-tech is deeply personal, originating from his training and practice as a medical doctor in Nigeria. There, Olatunji encountered firsthand the inefficiencies plaguing the system, notably the cumbersome paperwork process. He reflects on the frustration of repetitive tasks underserving human engagement, emphasizing the need for innovation to simplify routine processes, thus enabling medical professionals to focus on tasks of higher value.

Fuelled by these insights, Olatunji pursued further education in the U.S., earning degrees in medical informatics and computer science from the University of San Francisco and Georgia Tech, respectively. His career has spanned roles at notable tech firms, including Enlitic and Amazon Web Services, where he specialized in leveraging natural language processing (NLP) to enhance healthcare operations.

Such experiences have shaped his belief in the potential of U.S.-developed technologies to revolutionize healthcare in Nigeria and similar markets.

Intron Health emerged in 2020 with the goal of digitalizing Africa’s hospital systems through an Electronic Medical Record (EMR) System. Overcoming initial resistance from physicians who preferred handwriting, Olatunji pivoted towards enhancing data entry efforts. The discovery of prevalent transcription errors due to strong accents and complex medical terminology led to the development of Intron’s bespoke speech recognition tool, now deployed in 30 hospitals across five markets, including Kenya and Nigeria, making significant impacts such as drastically reducing waiting times for radiology results.

Olatunji highlights the critical nature of adopting new technologies in hospitals to leverage existing investments, revealing Intron’s role in facilitating the improvement of EMR system usage.

With a $1.6 million pre-seed funding round behind it, Intron Health is not only refining its technical capabilities, such as enhancing noise cancellation and supporting low-bandwidth operations, but also exploring the addition of decision support tools to minimize medical errors and improve patient care.

Intron stands among a rising tide of generative AI ventures in healthcare, like Microsoft’s DAX Express, striving to reduce clinicians’ administrative burdens. With the global market for speech and voice recognition projected to reach $84.97 billion by 2032, the innovation displayed by startups like Intron is critical. Furthermore, Intron is contributing to speech technology research in Africa, collaborating with institutions such as Google Research and the Bill & Melinda Gates Foundation to ensure the development of culturally nuanced models for use in local healthcare settings.