Using AutoLit to Complete a Review
The quality of a search is based on whether it includes all relevant studies (coverage) and how well it hones in on those studies without requiring screening of thousands of irrelevant studies. Our Search Optimization enables you to test search terms across indices (such as PubMed), with AI-enabled coverage estimates and predictions of how many studies you will include from each term.
Input sample search terms and screen several studies so that our software can learn what types of studies you are looking for. Use the suggested terms from our word clouds to further refine your search, and then finalize your search terms once you have reached the coverage percentage you are seeking.
Once you have optimized your search by testing sample terms, choose the final search terms and the indices you want results from. By running this search, you establish the set of studies that will be screened for inclusion in your review.
To import studies that were not available on the indices our search tool uses, upload a RIS file containing the metadata from the studies of interest and they will be automatically included in the screening process.
We offer several automatic screening methods, including exclusion of studies based on date of publication, MeSH tags, study type, or title and abstract text. Once you select the automated exclusions, you can craft custom exclusion criteria to fit the needs of your review.
Based on these custom exclusion criteria, review the studies that passed the automated exclusion filter and include them based on your discretion. As you do so, our software continuously updates its inclusion predictions and also presents the Patients, Interventions, Comparisons, and Outcomes (PICO) extracted from study abstracts to guide your screening decisions.
Our Study Tagging system allows you to build a review-specific, customized hierarchy of the data points relevant to your review. This serves two purposes: first, it enables you to build a disease-specific StudyViz to drill down on sub-populations, therapies, and outcomes of interest. Second, it establishes the data elements that you can extract from each study and teaches our platform the traits of that data element (is it continuous, ordinal, or binary? Is it an outcome? Is it reported as a mean or a median?).
By creating a tagging hierarchy and applying the tags you create to the studies in your review, you create a custom, updatable visualization of your search and pre-structure the data gathering that will be undertaken to create your comprehensive, updatable literature review.
Literature Search, Screening, and Tagging are the only steps needed to create a StudyViz
To publish your own StudyViz, as soon as you finish screening and tagging, just change the admin status of your StudyViz to make it public. Even if you keep it private, by completing Literature Search, Screening, and Tagging, you have automatically created your own personal StudyViz.
If you are completing a full literature review, the next step is to gather specific data points from studies, so proceed with Data Extraction!
Once you have established your tagging hierarchy, you should be able to rapidly create a structured dataset. The tags you applied can be used as data elements; on each study, define the treatment group and enter the study population; then, gather each datapoint from the studies of interest into our Data Extraction interface. If you have team members contributing to your work, use our Admin function to add them as data gatherers and track their progress.
We understand that export formats must be flexible to meet your needs. If you need a full description of methods, including PRISMA charts, full search protocols, and citations of each study included (and excluded) from your review, simply download a Word document that contains these methods as well as tables containing summary statistics and full raw datasets (in a file type of your choice).
In structuring these outputs, choose the type of review (Regulatory, Scientific, Competitive Intelligence/Marketing) to ensure that the methods and data are exported in the relevant format. If you would like to apply automated statistical analyses to your data or generate an interactive visualization that you can publish online, contact us for a consultation with our biostatistics team.