Intron Health receives support for its speech recognition tool that recognizes African accents

Voice recognition technology is becoming increasingly prevalent in various aspects of daily life. However, a significant issue persists: individuals who speak minority languages, as well as those with strong accents or speech impairments such as stuttering, often struggle to effectively utilize speech recognition systems for tasks like application control, transcription, and automation.

Tobi Olatunji, the founder and CEO of Intron Health—a clinical speech recognition startup—aims to address this challenge.

He asserts that Intron boasts Africa’s largest clinical database, with its algorithms developed using 3.5 million audio samples (equivalent to 16,000 hours) contributed by over 18,000 individuals from the healthcare field across 29 countries and representing 288 different accents.

Olatunji emphasizes that sourcing most contributors from healthcare professionals guarantees accurate pronunciation and capture of medical terminology for his intended audiences.

He mentioned that since we have already trained on a variety of African accents, it is highly probable that their access will perform significantly better than any other service they currently utilize.

He also noted that data collection from Ghana, Uganda, and South Africa is increasing, and the startup feels optimistic about implementing the model in those regions.

Olatunji’s passion for health technology arises from two key aspects of his background. Firstly, he received training and worked as a medical doctor in Nigeria, where he experienced firsthand the inefficiencies present in the healthcare system, such as the excessive paperwork involved and the challenges of managing it all effectively.

He said: Reflecting on my time as a doctor in Nigeria a few years back, I recall feeling quite frustrated with repetitive tasks that seemed unworthy of human effort, a sentiment I’ve held since my medical school days and continue to feel today. For instance, every lab order required me to write the patient’s name repeatedly. When I was seeing patients who needed prescriptions or lab tests, it became tedious to manually fill out each order with their name, age, date, and other details. This constant repetition left me wondering how we could improve our processes. I often found myself asking: What can we do to simplify things for doctors? Is there a way to delegate some of these tasks to another system so that physicians can focus more on valuable aspects of patient care?

Those inquiries pushed him into a new chapter of his life. Olatunji relocated to the United States to initially pursue a master’s degree in medical informatics at the University of San Francisco, followed by another in computer science at Georgia Tech.

He gained valuable experience working with several technology firms. As a clinical natural language processing (NLP) scientist and researcher at Enlitic, a company based in the San Francisco Bay Area, he developed models designed to automate the extraction of data from radiology reports.

Additionally, he worked as a machine learning scientist for Amazon Web Services. In both roles at Enlitic and Amazon, he concentrated on natural language processing applications in healthcare, contributing to systems that improve hospital operations.

During his various experiences, he began to contemplate how the innovations being utilized in the U.S. could enhance healthcare in Nigeria and similar developing markets.

Intron Health, which was established in 2020, initially aimed to modernize hospital operations across Africa by implementing an Electronic Medical Record (EMR) System. However, adoption proved difficult; according to Olatunji, many doctors favored writing over typing.

This realization prompted him to investigate ways to address this fundamental issue: improving the efficiency of physicians’ data entry and writing tasks.

Initially, the company considered external solutions for automating processes like note-taking and integrating existing speech-to-text technologies into their EMR system.

Numerous challenges arose due to frequent mis-transcription. Olatunji soon realized that the strong African accents and the complex pronunciation of medical terminology rendered existing foreign transcription tools ineffective.

This realization led to the development of Intron Health’s speech recognition technology, designed specifically to understand African accents and seamlessly integrate with current electronic medical records (EMRs).

So far, this tool has been implemented in 30 hospitals across five regions, including Kenya and Nigeria. The results have been encouraging; for instance, Olatunji noted that Intron Health has successfully cut down the waiting period for radiology results at one of West Africa’s largest hospitals from 48 hours to just 20 minutes.

Such improvements are vital for healthcare delivery in Africa, where the ratio of doctors to patients is among the lowest globally.

Hospitals have invested heavily in technology and equipment, making it crucial to utilize these advancements effectively.

We can offer valuable support to enhance the adoption of the EMR system, he noted. Looking forward, the startup is seeking new avenues for growth, having secured a $1.6 million pre-seed funding round led by Microtraction, with contributions from Plug and Play Ventures, Jaza Rift Ventures, Octopus Ventures, Africa Health Ventures, OpenseedVC, Pi Campus, Alumni Angel, Baker Bridge Capital, and various angel investors.

On the tech front, Intron Health is focused on refining noise cancellation technology while ensuring that their platform functions efficiently even with limited bandwidth.

Additionally, they are working on transcribing conversations with multiple speakers and incorporating text-to-speech features.

According to Olatunji, the strategy involves integrating intelligent systems or decision support tools for functions like prescribing medication and conducting lab tests.

He notes that these tools can minimize errors made by doctors, enhance patient care, and increase efficiency in their workflows.

Intron Health is part of a rising trend of generative AI startups in healthcare, such as Microsoft’s DAX Express, which are streamlining administrative responsibilities for healthcare providers by quickly generating notes.

The introduction and implementation of these technologies coincide with projections indicating that the global market for speech and voice recognition will reach $84.97 billion by 2032, growing at a compound annual growth rate (CAGR) of 23.7% starting in 2024, as reported by Fortune Business Insights.

In addition to developing voice technologies, Intron is significantly contributing to speech research in Africa through recent collaborations with Google Research, the Bill & Melinda Gates Foundation, and Digital Square at PATH.

This initiative aims to assess well-known Large Language Models (LLMs) like OpenAI’s GPT-4o, Google’s Gemini, and Anthropic’s Claude across 15 countries to pinpoint their strengths and weaknesses while addressing potential biases or risks. The ultimate goal is to provide culturally relevant models tailored for clinics and hospitals across Africa.