AI Tool Mia Partners with NHS, Identifies Critical Cancer Signs Overlooked by Clinicians
Early stage cancers can be difficult to spot with the human eye. Photo by Michael Hanschke/picture alliance via Getty Images.
Key Takeaways
- The UK’s National Health Service has successfully piloted an AI breast cancer screening tool, Mia.
- In a trial, Mia identified early stage cancers from mammograms more accurately than human doctors.
- Meanwhile, AI medical screening is also transforming the diagnosis of liver disease.
Detecting pathological anomalies in medical scans is an ideal use case for machine learning, which at heart, is all about analyzing data for patterns to make statistically informed predictions.
As more diagnostic centers deploy AI to help screen for diseases, the UK’s National Health Service (NHS) has successfully piloted Mia – a new tool that can identify signs of early stage breast cancer that are nearly invisible to the human eye.
Mia Delivers More Effective Cancer Screening
Early-stage cancerous growths can be extremely small and difficult to spot. As a result, as many as 20% of breast cancers are initially missed by mammogram screenings.
During a trial in which Mia analyzed the scans of over 10,000 women, not only did the AI tool successfully identify every cancer that had been flagged by human doctors, it also highlighted an additional 11 they missed.
By finding cancers earlier, Mia promises to deliver much better patient outcomes and could even save lives.
Because it works instantly, the tool’s developer, Kheiron, says it could the amount of time patients wait for results from an average of 14 days down to just three.
The Latest Frontier in AI Medical Screening
While Mia still has to be rolled out at scale, similar solutions for other types of disease screening have already delivered results.
Significant advances have been made in the use of AI to screen for different types of liver disease. Meanwhile, machine learning algorithms that process ultrasound data have performed better than human radiologists in identifying liver cancer.
As well as screening for illness, AI models can also help predict disease progression more effectively than traditional techniques.
Limitations in Medical Data
While AI diagnostic tools have the potential to significantly improve the efficiency and accuracy of medical screening, finding the right data to build properly working models remains a challenge.
For example, a 2022 study of AI liver disease screening tools found that, although they were highly effective at identifying pathologies in men, the tools missed 44% of cases in women.
In an interview with CCN, Unstoppable Domains COO and AI industry veteran Sandy Carter said the problem was the result of biased training data.
Carter said: “In the US, it wasn’t until 1993 that researchers had to include women in their health studies.”
But it isn’t just about improving women’s quality of life. Tools like Mia can also help the economy by reducing the impact of illness.
As Carter noted: “McKinsey just did a study . They found that if we could address this women’s health care gap, or the bias that exists in AI, it could boost the global economy by $1 trillion annually.”
UK Gov AI-Driven Civil Service Job Cuts, Shuns EU’s Specific AI Legislation
