Mining Google Trends data for nowcasting and forecasting colorectal cancer (CRC) prevalence

Area of Science:

• Oncology
• Public Health
• Data Science

Background:

• Colorectal cancer (CRC) is a leading cause of cancer death globally.
• Accurate CRC prevalence projections are crucial for resource allocation and strategy development.
• Limited global cancer statistics and high heterogeneity complicate CRC prevalence forecasting.

Purpose of the Study:

• To model, nowcast, and forecast global CRC prevalence.
• To evaluate Google Trends (GT) data as a surrogate for CRC survivor numbers.
• To identify reliable indicators for CRC disease prevalence.

Main Methods:

• A three-step framework using statistical and machine-learning models.
• Data mining to assess Google Trends (GT) relevancy for CRC prevalence.
• Utilizing ARIMA, ETS, and FNNAR models trained on GT time series (2004-2022).

Main Results:

• Population web-search interest in "colonoscopy" reliably indicates CRC prevalence.
• Colorectal cancer cases are predicted to rise over the next 24 months.
• A significant drop in CRC interest during the COVID-19 pandemic was observed.

Conclusions:

• Urgent public policies are needed to address modifiable CRC risk factors like lifestyle and nutrition.
• Public health authorities should enhance cancer screening rates during pandemics.
• Improved cancer statistics and global burden mitigation are potential positive outcomes.