A new feasibility study from Microsoft released in JAMA Oncology by Ryen White et al shows that it could be possible to screen for early signs of lung cancer by the web searches individuals perform.

Utilizing research data from Bing, researchers found that careful analysis of a patient’s search history could be used as a pre-screening tool for lung cancer.

Anonymous searches on Bing were scanned for indicators of a recent lung cancer diagnosis: searches for symptoms, first-person phrases, and more. For example: Their example is “I was just diagnosed with lung cancer,” and follow-up searches for treatment. They also reviewed demographics, location, and red flags like searches for smoking cessation products. The researchers then looked at what that individual searched for over the previous year, specifically looking for queries of symptoms like chest pain, cough, bronchitis and found that these queries correlated with an increased chance of lung cancer diagnosis.

For cohorts exhibiting risk factors, this method “makes accurate predictions up to 1 year prior to experiential diagnostic queries (eg, detecting 10% of positives while being incorrect 1 in 10,000 times).” But the study concedes that information from the patient such as family history, age and smoking habits would need to be gleaned directly from the searcher.

The findings may prove to be useful for doctors and patients pre-diagnosis. Or, the researchers posit, it could be used to warn those with concerning web search phrases to discuss screening for lung cancer sooner. But the researchers concluded that this might cause undue panic and unnecessary healthcare costs — which in itself could cause serious patient harm.

Source: Evaluation of the Feasibility of Screening Patients for Early Signs of Lung Carcinoma in Web Search Logs