Back

Accelerating Insight Discovery in Large Biomedical Text with Scalable Processing Framework

Kim, D.; Hauptman, M.; Patrick, M. T.

2025-08-19 bioinformatics
10.1101/2025.08.14.670384 bioRxiv
Show abstract

Large language models are increasingly being used by dermatology professionals to support diagnostic investigation, patient education, and medical research. While these models can help manage information overload and improve efficiency, concerns persist regarding their accuracy and potential reliance on dubious sources. We introduce Quanta, a hybrid system that combines large language models with established evaluation metrics, such as cosine similarity, to enable efficient summarization and interpretation of curated research corpora. This methodology ensures that synthesized insights remain domain-specific and contextually relevant, thereby supporting clinicians and researchers in navigating the expanding digital landscape of dermatology literature. Deployed within an interactive chatbot, the tool delivers direct answers to user queries, provides cross-publication insights, and can suggest new directions for research. Comparative evaluations on benchmark datasets demonstrate improvements in accuracy, efficiency, and computational cost, with the curated document approach enhancing reliability and reducing misinformation risk.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

1
Bioinformatics
◐ 1204 papers in training set
Top 1%
18.6%
Similar papers in this journal
2
Bioinformatics Advances
◐ 203 papers in training set
Top 0.1%
12.0%
Similar papers in this journal
3
GigaScience
◐ 212 papers in training set
Top 0.2%
9.0%
Similar papers in this journal
4
Computational and Structural Biotechnology Journal
● 242 papers in training set
Top 0.3%
6.8%
Similar papers in this journal
5
Patterns
○ 78 papers in training set
Top 0.3%
4.9%
Similar papers in this journal
50% of probability mass above
6
Journal of Medical Internet Research
◐ 87 papers in training set
Top 0.5%
4.3%
7
BioData Mining
○ 22 papers in training set
Top 0.1%
4.1%
8
Scientific Reports
○ 3612 papers in training set
Top 31%
3.3%
9
Database
◐ 61 papers in training set
Top 0.3%
2.7%
10
BMC Bioinformatics
○ 457 papers in training set
Top 3%
2.4%
11
Journal of the American Medical Informatics Association
◐ 71 papers in training set
Top 1%
2.1%
12
PLOS Digital Health
● 106 papers in training set
Top 2%
2.1%
13
PLOS ONE
● 5266 papers in training set
Top 46%
1.7%
14
Journal of Biomedical Informatics
○ 47 papers in training set
Top 0.8%
1.7%
15
Briefings in Bioinformatics
◐ 354 papers in training set
Top 6%
1.1%
16
PLOS Computational Biology
● 1863 papers in training set
Top 17%
1.1%
17
Frontiers in Genetics
○ 230 papers in training set
Top 4%
1.1%
18
Artificial Intelligence in Medicine
○ 17 papers in training set
Top 0.5%
1.1%
19
Computers in Biology and Medicine
○ 128 papers in training set
Top 4%
1.1%
20
BMC Medical Informatics and Decision Making
○ 43 papers in training set
Top 1%
1.0%
21
Genome Biology
○ 637 papers in training set
Top 8%
1.0%
22
iScience
○ 1154 papers in training set
Top 32%
1.0%
23
Frontiers in Bioinformatics
○ 49 papers in training set
Top 1%
1.0%
24
IEEE Access
● 35 papers in training set
Top 1%
0.8%
25
JMIR Medical Informatics
◐ 18 papers in training set
Top 0.8%
0.8%
26
Journal of Chemical Information and Modeling
● 238 papers in training set
Top 2%
0.6%
27
Nucleic Acids Research
◐ 1281 papers in training set
Top 14%
0.6%
28
Nature Communications
○ 5641 papers in training set
Top 58%
0.6%

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.