Lit Scout¶

Client: Dr Mike Trower, White Wings Consulting.

The client is a former founder/CSO in biotech drug discovery and now consults as White Wings Consulting. Researchers like him need to keep track of new scientific papers and patents in their interest areas, but manually searching hundreds of references across public databases such as PubMed, PubChem, and patent sites like Espacenet every few weeks is time-consuming and easy to get wrong, especially since many relevant patents are filed in foreign languages such as Chinese and never turn up in an English keyword search. Your task is to build a tool that runs a scheduled search across these databases using a keyword set (eg compound names, protein targets, disease terms), automatically translating foreign-language patents so they can be searched too, and then filters results down to only those that (a) match a defined sub-topic list, and (b) have been published since the last scheduled run. The tool should sort surviving results by topic and study type (eg preclinical, clinical, review), skip anything already flagged previously, and produce a dated table that includes an explanation of why each item was included. The hard part is building translation and relevance-screening that stays reliable across very different databases, document formats, and languages.