diff --git a/StudentNotebooks/Assignment07/liux30_assignment07.Rmd b/StudentNotebooks/Assignment07/liux30_assignment07.Rmd index f900113..9ab4fd6 100644 --- a/StudentNotebooks/Assignment07/liux30_assignment07.Rmd +++ b/StudentNotebooks/Assignment07/liux30_assignment07.Rmd @@ -4,15 +4,18 @@ author: "Xiheng Liu" date: "`r Sys.Date()`" output: pdf_document: - toc: yes + toc: true toc_depth: '3' html_notebook: default html_document: - toc: yes + toc: true toc_depth: 3 - toc_float: yes - number_sections: yes + toc_float: true + number_sections: true theme: united + word_document: + toc: true + toc_depth: '3' --- # DAR Project and Group Members @@ -22,18 +25,28 @@ output: -# 0.0 Preliminaries. +# 0.0 Preliminaries This report is generated from an R Markdown file that includes all the R code necessary to produce the results described and embedded in the report. # 1.0 Project Introduction +DAR-CTEval-F24 evaluates large language models (LLMs) in clinical trial design, focusing on their ability to suggest and evaluate baseline patient features using real trial data. The benchmark measures accuracy and performance by comparing LLM-generated results with actual trial features and human evaluations. -# 2.0 Organization of Report +CTBS, a component of DAR-CTEval-F24, is a specialized tool tailored for clinical trial design. Building upon the ChatBS framework, it serves as a comprehensive benchmark for assessing language model capabilities in this domain. +This report provides a programming guide for the CTBS App. -This report is organize as follows: +Link to Project Github: https://github.rpi.edu/DataINCITE/DAR-CTEval-F24 + +Link to CTBS Github: https://github.rpi.edu/DataINCITE/DAR-CTBSApp-F24 +Link to CTBS App: https://lp01.idea.rpi.edu/shiny/erickj4/DAR-CTBSApp-F24/ + + +# 2.0 Organization of Report + +This report is organize as follows: * Section 3.0 Page 1: 'Step 1: Specify Trial' is divided into left and right sides. There is a button on the left side. After pressing the left button, the right side will change. @@ -48,6 +61,8 @@ This report is organize as follows: # 3.0 Page 1: 'Step 1: Specify Trial' +In 'Step 1: Specify Trial', user can query the Title, Brief Summary, Condition, Eligibility Criteria, Intervention, Outcome, Actual Features of a specific NCT ID in CT_Pub and download it as a JSON document. + The relevant code for this page is as follows: The code for UI part: line 192-255 @@ -59,13 +74,20 @@ The code for download in server part: line 1243-1286 ## 3.1 Data, Code, and Resources +Here is a list data sets, codes, that are used in CTBS. + +1. liux30_assignment07.Rmd (with knit pdf and html) is this notebook. +[https://github.rpi.edu/DataINCITE/DAR-CTEval-F24/blob/main/StudentNotebooks/Assignment07/liux30_assignment07.Rmd](https://github.rpi.edu/DataINCITE/DAR-CTEval-F24/blob/main/StudentNotebooks/Assignment07/liux30_assignment07.Rmd) + ## 3.2 Contribution +This page is completed by Xiheng Liu and Tianyan Lin. ## 3.3 Methods Description +I created an observeEvent function for the ‘Load’ button. After clicking the button, it will first determine whether the input NCTid is in the database. If not, it will return ‘Cannot find this NCTId.’. If it is, the row where the NCTid is located will be extracted, and Title, BriefSummary, Condition, EligibilityCriteria, Intervention, and Outcome will be stored separately according to the column names. Then put it in the user prompt template that starts with ‘##Question:’ and ends with ‘##Answer:’. It will then be transmitted to LLM. ## 3.4 Result and Discussion @@ -75,7 +97,6 @@ The code for download in server part: line 1243-1286 - # 4.0 Page 2: 'Step 2: Generate Descriptors' The relevant code for this page is as follows: @@ -88,10 +109,22 @@ The code for output in server part: line 942-952, 966-990 ## 4.1 Data, Code, and Resources +Here is a list data sets, codes, that are used in CTBS. + +1. liux30_assignment07.Rmd (with knit pdf and html) is this notebook. +[https://github.rpi.edu/DataINCITE/DAR-CTEval-F24/blob/main/StudentNotebooks/Assignment07/liux30_assignment07.Rmd](https://github.rpi.edu/DataINCITE/DAR-CTEval-F24/blob/main/StudentNotebooks/Assignment07/liux30_assignment07.Rmd) + + ## 4.2 Contribution +This page is completed by Xiheng Liu and Tianyan Lin. + ## 4.3 Methods Description +For Submit button, we have zero-shot and Three-shot. + +For Three-shot function, we should take three trials as example, so I selected the example rows and target row for the ids we want. Then save all the elements for defferent columns to build the user propmt. + ## 4.4 Result and Discussion ## 4.5 Conclusions, Limitations, and Future Work. @@ -111,10 +144,21 @@ The code for downlaod in server part: line 1288-1317 ## 5.1 Data, Code, and Resources +Here is a list data sets, codes, that are used in CTBS. + +1. liux30_assignment07.Rmd (with knit pdf and html) is this notebook. +[https://github.rpi.edu/DataINCITE/DAR-CTEval-F24/blob/main/StudentNotebooks/Assignment07/liux30_assignment07.Rmd](https://github.rpi.edu/DataINCITE/DAR-CTEval-F24/blob/main/StudentNotebooks/Assignment07/liux30_assignment07.Rmd) + + ## 5.2 Contribution +This page is completed by Samuel Park and Xiheng Liu. + ## 5.3 Methods Description +For Evaluation function, we should get all elemens in input textboxs, and also get the ActualFeatures and CandidateFeatures. The ActualFeatures is the Paper_BaselineMeasures_Corrected in CT_Pub_updated.df, and we save it into input box. After run load button, we get the result, and we can use regular expression to get the content between the first "]" and the first "*", which is CandidateFeatures. Then also save to input box. I build the user propmt bases on all the element in input boxes. + + ## 5.4 Result and Discussion ## 5.5 Conclusions, Limitations, and Future Work. diff --git a/StudentNotebooks/Assignment07/liux30_assignment07.pdf b/StudentNotebooks/Assignment07/liux30_assignment07.pdf index df5fb93..34e4db1 100644 Binary files a/StudentNotebooks/Assignment07/liux30_assignment07.pdf and b/StudentNotebooks/Assignment07/liux30_assignment07.pdf differ