MIT researchers develop AI model to detect future lung cancer risk

A team of researchers from the Massachusetts Institute of Technology (MIT) has developed an artificial intelligence (AI) model that is able to detect future lung cancer risk.

The assessment tool, known as Sybil, is being developed by researchers at MIT's Abdul Latif Jameel Clinic for Machine Learning in Health, Mass General Cancer Center (MGCC) and Chang Gung Memorial Hospital (CGMH).

Lung cancer is the leading cause of cancer death in the US, accounting for about one in five of all cancer deaths.

If the disease is found at an early stage, when it is small and before it has spread, it is more likely to be treated successfully. However, symptoms of lung cancer usually do not appear until the disease is already at an advanced stage.

Currently, low-dose computed tomography (LDCT) scans of the lung are the most common way patients are screened for the disease.

The US Preventive Service Task Force recommends annual LDCTs for those aged over 50 to 80 years of age with a 20 pack-year history of smoking, who either currently smoke or have quit smoking within the last 15 years. However, less than 10% of eligible patients are screened annually.

Sybil builds on the screening process further, analysing the LDCT image data without the assistance of a radiologist to predict the risk of a patient developing a future lung cancer within six years.

The study, published in the Journal of Clinical Oncology, details how the AI model was validated on three independent data sets: a set from over 6,000 LDCTs from the National Lung Screening Trial (NLST), over 8,000 LDCTs from Massachusetts General Hospital and over 12,000 from CGMH.

The tool was able to accurately predict the risk of lung cancer across all three sets. The team determined Sybil’s accuracy by using ‘area under the curve’ (AUC), a measure of how well a test can distinguish between disease and normal samples.

With 1.0 as the highest possible score, Sybil predicted cancer within one year with AUCs of 0.92 for the NLST participants, 0.86 for the Massachusetts General Hospital data set and 0.94 for the data set from CGMH.

The programme predicted lung cancer within six years with AUCs of 0.75, 0.81 and 0.80 respectively, for the three data sets.

The team concluded that Sybil can accurately predict an individual's future lung cancer risk from a single LDCT scan to further enable personalised screening, but noted that further studies are required to understand the tool’s clinical applications.

Digital inspiration for pharma on mobile, social media, strategy, best practice, regulations and more