This workflow uses the Text Chunker node to process and summarize lengthy medical research papers. The Text Chunker node is essential due to the large document size, which cannot be processed in a single pass.
You can download and run the workflow directly in your KNIME Analytics Platform. For optimal performance, we recommend using the latest version of KNIME AP. This workflow can also be deployed as a data application in KNIME Business Hub.
Workflow steps
Paper Configuration
Click the lens icon to execute the first component, "Paper Configuration." Select the paper to summarize, generate notes, or create flashcards. Adjust the chunk size and overlap settings as needed.
It is crucial to set an appropriate overlap to avoid losing context between chunks.
Chunk Processing
After chunking, we loop over each chunk and pass it to the LLM Prompter node. The prompt's initial message is critical: "You are a senior medical researcher. You receive chunks of a medical research paper as input and are expected to help summarize the entire paper. There is some overlap between the chunks to avoid losing context." + Chunk.
Summarizing Chunks
After the LLM Prompter is executed we will have multiple chunk summaries. This historical conversation data is passed to the next Chat Model Prompter node to produce comprehensive summaries, notes, or flashcards.
The system message inside the node reiterates: "You are a senior medical researcher. You receive chunks of a medical research paper as input and are expected to help summarize the entire paper. There is some overlap between the chunks to avoid losing context."
Generating Final Output
Based on the selected output type, the last Chat Model Prompter node message will be, e.g., "I am giving you all the summaries produced from the entire paper. Take them and produce a comprehensive summary of the whole paper."
The response from the assistant is then displayed in a Table View node.
Papers used in this workflow:
Naciri A, Radid M, Kharbach A, Chemsi G. E-learning in health professions education during the COVID-19 pandemic: a systematic review. J Educ Eval Health Prof. 2021;18:27. doi: 10.3352/jeehp.2021.18.27. Epub 2021 Oct 29. PMID: 34710319; PMCID: PMC8609102.
Lin, YK., Saragih, I.D., Lin, CJ. et al. Global prevalence of anxiety and depression among medical students during the COVID-19 pandemic: a systematic review and meta-analysis. BMC Psychol 12, 338 (2024). https://doi.org/10.1186/s40359-024-01838-y
Harsch IA, Konturek PC. The Role of Gut Microbiota in Obesity and Type 2 and Type 1 Diabetes Mellitus: New Insights into "Old" Diseases. Med Sci (Basel). 2018 Apr 17;6(2):32. doi: 10.3390/medsci6020032. PMID: 29673211; PMCID: PMC6024804.
Khalaf AM, Alubied AA, Khalaf AM, Rifaey AA. The Impact of Social Media on the Mental Health of Adolescents and Young Adults: A Systematic Review. Cureus. 2023 Aug 5;15(8):e42990. doi: 10.7759/cureus.42990. PMID: 37671234; PMCID: PMC10476631.
Durmuş Şenyapar, H. Nurgül. (2023). Benefits of Renewable Energy for Public Health: A Bibliometric Analysis. Journal of Energy Systems. 7. 10.30521/jes.1252122.