Automatic patient pre-screening for clinical trials: a narrative review

ACL ARR 2025 July Submission158 Authors

Abstract: We present a narrative review of recent advances in Natural Language Processing for automating patient pre-screening in clinical trials. We review the state-of-the-art across three core tasks: (1) automatic generation of eligibility surveys from trial protocols, (2) extraction of structured patient information from electronic health records (EHRs) and (3) automatic patient-trial matching. We analyze recent trends in using neural architectures, and we highlight current bottlenecks in linguistic variability, data interoperability and hallucination in generative systems. Our survey aims to synthesize a fragmented landscape and provide future directions towards clinical trials improvement.

Paper Type: Long

Research Area: NLP Applications

Research Area Keywords: question generation,clinical NLP,LLM/AI agents

Contribution Types: Surveys

Languages Studied: English,Spanish

Reassignment Request Area Chair: This is not a resubmission

Reassignment Request Reviewers: This is not a resubmission

A1 Limitations Section: This paper has a limitations section.

A2 Potential Risks: No

A2 Elaboration: As a survey, there is no direct potential risks of our work

B Use Or Create Scientific Artifacts: Yes

B1 Cite Creators Of Artifacts: Yes

B1 Elaboration: Appendix B Table 5, we extracted criteria from a publicly available clinical trial

B2 Discuss The License For Artifacts: No

B2 Elaboration: Appendix B Table 5, the only artifact used comes from publicly available sources

B3 Artifact Use Consistent With Intended Use: No

B3 Elaboration: As a survey, most of the work was literature review, not experimental setup

B4 Data Contains Personally Identifying Info Or Offensive Content: No

B4 Elaboration: As a survey, most of the work was literature review, not experimental setup

B5 Documentation Of Artifacts: Yes

B5 Elaboration: Section 5.3.2

B6 Statistics For Data: Yes

B6 Elaboration: Appendix B table 3

C Computational Experiments: Yes

C1 Model Size And Budget: No

C1 Elaboration: As a survey, most of the work was literature review, not experimental setup

C2 Experimental Setup And Hyperparameters: No

C2 Elaboration: As a survey, most of the work was literature review, not experimental setup

C3 Descriptive Statistics: No

C3 Elaboration: As a survey, most of the work was literature review, not experimental setup

C4 Parameters For Packages: No

C4 Elaboration: As a survey, most of the work was literature review, not experimental setup

D Human Subjects Including Annotators: No

D1 Instructions Given To Participants: No

D1 Elaboration: As a survey, most of the work was literature review, not experimental setup

D2 Recruitment And Payment: No

D2 Elaboration: As a survey, most of the work was literature review, not experimental setup

D3 Data Consent: No

D3 Elaboration: As a survey, most of the work was literature review, not experimental setup

D4 Ethics Review Board Approval: No

D4 Elaboration: As a survey, most of the work was literature review, not experimental setup

D5 Characteristics Of Annotators: No

D5 Elaboration: As a survey, most of the work was literature review, not experimental setup

E Ai Assistants In Research Or Writing: Yes

E1 Information About Use Of Ai Assistants: Yes

E1 Elaboration: Section 3

Author Submission Checklist: yes

Submission Number: 158