Managing Director, Corporate Health International
AI and Augmented Humanity
CorporateHealth International (CHI) provides services to build and deliver tele-endoscopy investigations of the lower gastrointestinal (GI) tract. The goal is that eventually, every person with a gastrointestinal disease should have the chance to get diagnosed early and people’s risk of colorectal cancer (CRC) is accurately predicted without the need for highly invasive procedures.
Diseases of the GI from Crohn’s to colorectal cancer have some of the most harmful and widespread impact on the health of people as well as on the healthcare resources of a country. For more than 63 years, undertaking an endoscopy was the only way to accurately diagnose the state of a patient’s digestive tract – an invasive medical procedure is stressful, painful and risky. Some patients must travel for miles to reach a hospital and in many countries waiting lists are long.
During the pandemic elective procedures were paused and now capacity is further reduced by up to 75% due to additional cleaning requirements to address the increased risk of infection through aerosolized particles of the GI tract. This means that patients cannot be seen in time and cancers will be missed. Adding to this, the procedure requires multiple staff and complex equipment in dedicated facilities and most of the time the patient gets sedated, leading to a lengthy procedure. The alternative procedure for the diagnostic process is to visually inspect the bowel with a minimally invasive device, a video capsule endoscope.
CHI developed a comprehensive managed service that takes care of all the steps to conduct a video capsule investigation, the first of its kind implemented in Scotland.
The process involves the patient wearing a hip-strapped data recorder to capture up to 400,000 wirelessly transmitted images. Once the capsule is swallowed, the patient should stay active, but can do that at home or any other location. This highlights the other main advantage of the solution: The procedure can be conducted in the patient’s community, possibly even in their home - away from pandemic risks.
The recorder then uploads these images to two Intel ® Xeon® Scalable processor-based servers that analyse the data. AI algorithms then quickly mark-up abnormalities which has drastically reduced the time to process one video from 2 hours to 20 minutes. Human experts maintain full control of the analysis. Intel plays a tech advisory role, ensuring patient data meets privacy and regulatory controls.
The managed service provides data as accurately as any endoscopy, minus the stigma, fear, and at continuously decreasing costs. The NHS coined it ‘SCOTCAP’ – Scottish Capsule Program. After 7 sites across the northwest of Scotland took the lead to conduct a service evaluation of over 500 patients, the Scottish government is driving the roll-out across the whole country, supporting the need for remote treatment as a result of the pandemic. This solution is also being scaled in England, with the aim to reach other European and North American countries in the future.
The AI development of this solution was supported by Intel - preparing datasets for machine learning and machine learning optimizations. The Intel team set-up high-performance data processing systems as well as the servers to host the neural networks to be trained, which would have taken CHI many days and even weeks to do on their own.
The development of the managed solution resulted in many measured improvements compared to the optical colonoscopy procedure. The following are true about the managed solution:
- It was the first new radical endoscopy procedure in 63 years
- 400,000 images taken of each patients intestinal lining.
- Processing time to analyse reduced from 2 hours to 20 minutes.
- It’s being rolled out across 14 health boards and multiple English trusts
- Accuracy is on par with optical colonoscopy
- The costs of the procedure are currently comparable to traditional colonoscopy.
- Reduces patient time from 2-5 hours (with/without sedation and travel) to 30 minutes
- Medtronic-built colon capsule: PillCam Colon II
- Newly developed ML learning algorithms to address small number of existing datasets in collaboration with University of Barcelona.
- Custom-built data set preparation tools to accelerate ML-data creation in collaboration with Wolfram Technologies.
- Intel® Xeon® Scalable processors which deliver industry leading, workload optimized performance, with built-in AI acceleration.
CorporateHealth International have transformed the whole endoscopy procedure, from patient experience to analysing results. Intel have enhanced this process further by using AI-algorithms, resulting in drastic time savings for doctors and nurses.
CHI is innovating further, within two years they aim to remove the need for specialised nurses, and within four years the doctors that are currently required to perform the analysis, saving a huge amount of time which can then be spent on patient care.
This solution is now being rolled out in 100 hospitals in Scotland; up to 10,000 symptomatic patients will be able to receive an accurate bowel cancer risk assessment in the next 9 months and the cases will grow across the UK. It allows us to continue finding cancers with minimal disruption during the pandemic, which sheds a light on the flaws of the status quo and highlights the opportunity to use a more patient friendly, less risky and more efficient solution that can be rolled out quickly.