## If you are working in healthcare and in AI, then course like these are essentials. will help to navigate complexities of "Clinical datasets"! (Even GenAI will require this knowledge)
1- CLINICAL Named Entity Recognition - clinical entities have several variability than general entities and can be from fields like medication, disease, symptoms, ADE, anatomy, duration of medication, gene, lab test, route of administration. Framework like scispacy, CliNER, CLAMP, cTakes are good for Clinical NER.
Practical uses. - extraction of entities such as genes/biomarkers, create knowledge base, pattern analysis, detection of medication names.
2. Clinical Entity Resolution - Extraction of entities are good, there are lot to be done in this space for example linkage of disparate clinical data sources, disambiguation, deduplication of clinical entities for data quality. And this is where clinical databases like UMLS, MeSH (Medical subject headings), RxNorm, Go (Gene Ontology), HPO (Human Phenotype Ontology)
3. Clinical Text Representation - clinical text like case note, family history, lab reports are some of most used documents in clinical world. They have abbreviation, notation and several domain nuances as well.
Different CLINICAL Transformers can be put to use here like
biobert, pubmedbert, bert-clinicalQA, clinical-longformer, clinical-bigbird, roberta-base-biomedical, Core clinical diagnosis prediction
Some of transformers like bert-base-uncased-clinical-ner, biomedical-ner can be used to detect clinical entities as well.
#generativeai #ai #ushealthcare #clinicalnlp
=======================================================
Just finished the course “Advanced AI: NLP Techniques for Clinical Datasets” by Wuraola Oyewusi! Check it out: