Sample Files: AI-Complaint-Assistant with prompt
You can share your own app files on this page. Please include a brief description of your app files.
Reference files File:ComplaintAssessmentPrompt-Schema.zip File:OpenFDA device search fields.xlsx
- Settings.txt
- CONFIG:OpenFDA-Vertex-V1.0 //identifies schema version
- Open FDA API Query specification:
- https_url1:api.fda.gov/device/event.json?
- limit1:100
- REPORT
- report_title:Complaint Reportability Assessment
- // Data fields analyzed in MDR query results
- data1_fields:product_problems, device.brand_name, patient.patient_problems, device.manufacturer_d_name, report_number, mdr_text.text
- // General keyword search terms (in any field)
- KEYWORD1-SEARCH:
- // Query search fields and terms
- SEARCH1-FIELDS:SEARCH1-TERMS
- date_received:[20200101+TO+20240315]
- device.device_report_product_code:(FMF)
- // Query results sorting
- SORT1-FIELD:SORT1-TERM
- date_received:desc
- // Query count field/term
- COUNT1-FIELD:COUNT1-TERM
- AI CONFIGURATION & PROMPTS:
- // Prompt that summarizes the product problem
- AI-ProblemSummaryPrompt:Describe the following product problem in a couple of sentences. Include essential details.
- // Prompt that counts the instances of items
- AI-CountSummaryPrompt:List each unique item along with the count for each item.
- // Prompt that summarizes the problem similarity to a product problem
- AI-ProblemSimilarityPrompt:Analyze the similarity of this problem (Template:Problem input) to the following problem. Use semantic similarity measurement. Present the result with the similarity score as a percentage followed by a concise explanation of the similarity score.
- // Prompt that analyzes the similarity between input problem and MAUDE results matching query criteria
- AI-MDRSimilarityPrompt:Match the most similar problem reports to this (Template:Problem input). Include all important details. Include all reference numbers. Include the similarity scores as a percentage. Explain the similarities. Use semantic similarity measurements. Present results in descending similarity scores. For each matching result, include the matching report number followed by the similarity score followed by a brief description of the problem followed by an explanation of the similarities.
- // Prompt that summarizes the most similar problems
- AI-ReportSummaryPrompt: List the top matching problem reports with the highest similarity scores. Include all details about the matching problems.
- // Maximum number of words in each intermediate report.
- AI-WordsPerReport:1500
- // LLM Pro-Vision temperature index (0..1f)
- AI-ModelTemperature:0.05
- // LLM Pro-Vision TOP_P index (0..1f)
- AI-ModelTopP:0.4
- // LLM Pro-Vision TOP_K (Number of words for next word prediction)
- AI-ModelTopK:10
- // LLM Maximum output words (1..2048)
- AI-ModelMaxOutputTokens:2048
Sample App Files
Media: Basic-Complaint-Assessment-with-prompt.zip Description: Settings includes filter on product code FMF and date range from 2020 to present. Prompts for matching top similar product problem codes and individual MAUDE reports using semantic similarity.