AIDS Search Trends and Reported Cases in China 2011-2017

78 months of time-series data from January 2011 to June 2017 combine Baidu search trends with officially reported AIDS occurrences and deaths. The dataset supports a machine learning method, artificial neural networks, used to forecast AIDS dynamics. It was created by Yongqing Nan and published in 2020.

Use Cases

• Train an artificial neural network model using Baidu search trend data as input variables to forecast officially reported AIDS occurrences.
• Analyze the correlation between specific AIDS-related search query volumes from Baidu and the output variable of monthly AIDS deaths.
• Validate forecasting performance using the mean absolute percentage error, root mean squared percentage error, and index of agreement metrics on the 78-month time series.

Strengths

• Provides 78 months of continuous time-series data from January 2011 to June 2017.
• Integrates two distinct data sources: search engine query trends and official public health reports.
• The associated research paper validates the forecasting method using three specific statistical criteria.

Limitations

• The specific search queries used as input variables and the raw data rows are not detailed.
• Data is temporally stale, ending in mid-2017 and not reflecting recent trends.
• Geographic scope is limited to China, limiting applicability to other regions.

Provenance

Source

Dryad digital repository.

Collection Method

Search trend data collected from Baidu.com; official AIDS occurrence and death data from Chinese health authorities.

Time Range

January 2011 to June 2017.

Freshness

Last updated in June 2020, with underlying data ending in June 2017.

Geography

China.