I am searching for an aid that helps to find the most relevant intervention studies for a given person's (bio)medical profile and intervention objective.
This aid should apply NLP or ACA techniques to extract from publication resources (e.g. NLM) (a) profile information (inclusion and exclusion criteria) of study participants, (b) type of intervention, (c) effect.
My extensive literature search has not thrown up any existing system that meets this description. I'd be grateful for relevant information, either about such systems or about individuals/groups with the necessary expertise/experience to develop such a system.
Not sure if this could be of use, but I have developed a Language Engine (Enguage) which is entirely configured/programmed in Natural Language allowing people to interact with software through NL. It's probably not what you're envisioning in that it doesn't (yet) read and understand (essentially unstructured) written text, as "profiles of study participants probably" will be. But it does let you interact with software (and by extension hardware) through voice. So it could be used for data entry in by voice. More details and links to video at www.enguage.org - or go direct to YouTube and search for my name and/or Enguage. All my published papers are here on Rg.
Kind regards, Martin.
Not sure if this correctly addresses your question, but Massive Bio SYNERGY-AI program matches cancer patients with active clinical trials, you can find more info here https://massivebio.com/find-a-clinical-trial/free-clinical-trial-patient-matching it's a promising solution in my view.
Dear Christine, Thanks for the link. That is a very good match, indeed.
The only difference between Massive Bio and my intended application: Massive Bio looks for registered trials. These are all registered in public trial registries which systematically record all relevant participant profile data (inclusion and exclusion criteria), and, of course, the intervention profiles of the trial. It's relatively easy to find a match within the registries' structured databases for a given individual patient's profile.
Contrary to that, my intention is to find the best match for my patients' profiles among trials that have been completed already, and to extract the relevant results from there. Those completed trials may also have been registered in a trial registry, but the farther you go back in time, the less likely this is the case. Because only since the last couple of years have the journals begun to require trial registration as a prerequisite for publishing papers that present trial results. And the content of papers is a lot less structured than the files of the registries.
If you are aware of any existing solution to my particular problem, I'd be grateful for a link. And, if you are interested in helping develop such a solution under a potential grant project, I'd be happy to brief you on the detailed requirements.
Cheers
Dear Prof. Lutz Kraushaar , thanks for clarifying, I see your need, but unfortunately I am not aware of such solution, will sure let you know if I come across something similar.
And thanks for the collaboration offer, this is indeed a very interesting project, my expertise is not on the technical development side as much as it's on the user adoption /organisational implications side... if this kind of expertise is needed for this project, I'd sure be interested to know more.
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication