Research led by the University of Essex and Check4Cancer has developed a new AI framework that can detect 85 percent of skin cancer when combined with some of the existing assessment metrics. Researchers assert that “AI is set to revolutionize the way skin cancer can be detected.”
Malignant melanoma is the 17th most common cancer worldwide and is solely responsible for 80% of all skin cancer deaths. Reports estimate that a delay in the detection of Melananoma reduces the 5-year survival rate by 20%.
The increasing population and COVID-19 have significantly contributed to increasing healthcare access pressure and challenges in delivering timely assessments and diagnoses.
Therefore, the study urges for new methods for cost-effective and secured skin cancer detection; “There is a need to develop new methods that can be used as a decision aid for the classification of suspicious or non-suspicious skin lesions during teledermatology triage.”
The new AI framework, known as the C4C risk score, is devised with skin lesion metadata collection, identification of a new list of skin cancer risk factors, and proposal of a new risk score. C4C considers 5 AI models to classify skin lesions into suspicious and non-suspicious classes accurately.
The study published in Nature Scientific Reports analyzed collections of over 53,601 skin lesions from 25,105 patients. Using deep learning and machine learning, the C4C risk score outperforms the existing methods (7-point checklist and Williams method), with a balanced accuracy of 71%.
Apart from the 22 meta-features, the framework considers age, gender, hair color, and family history to conclude whether a lesion is malignant. Additionally, the AI also evaluates the data from 450,000 patients between 1997 and 2015.
New diagnostic tool detects earliest signs of heart attack in minutes
“We have developed an AI framework solely based on metadata and observed that it can separate suspicious skin lesions from non-suspicious ones with high sensitivity. The C4C risk score can be used as a decision-aid by telemedicine reporters to help with final lesion classification that is equivocal after image classification alone. This has the potential to reduce the number of referrals to a special clinic for possible biopsy and help reduce the waiting times for skin cancer diagnosis,” says the lead researcher Dr Haider Raza.
While AI-based skin cancer detection has demonstrated improved performance, detection solely based on metadata remains questionable.
Therefore, scientists are moving forward to the stage of adding image assessment methods to detect C4C risk scores accurately.
Interestingly, researchers are developing a C4C smartphone app for quick and easy assessment of lesions.
The electric field generating stitches speed up healing by 50%
Journal Reference
- Islam, S., Wishart, G. C., Walls, J., Hall, P., G., A., Gan, J. Q., & Raza, H. (2024). Leveraging AI and patient metadata to develop a novel risk score for skin cancer detection. Scientific Reports, 14(1), 1-12. DOI: 10.1038/s41598-024-71244-2
