From e29c74ac02677104c5f057895e72302b66144bc1 Mon Sep 17 00:00:00 2001
From: isc-lperezra <lperezra@intersystems.com>
Date: Tue, 24 Dec 2024 15:24:23 +0100
Subject: [PATCH] LLM selection available

---
 .../src/app/analyzer/analyzer.component.html  |  4 ++-
 .../src/app/analyzer/analyzer.component.scss  |  4 +++
 .../src/app/analyzer/analyzer.component.ts    | 28 ++++++++++++++-----
 encoder-ui/src/assets/i18n/en.json            |  3 +-
 encoder-ui/src/assets/i18n/es.json            |  3 +-
 src/ENCODER/BP/AnalyzeTextProcess.cls         | 13 ++++-----
 src/ENCODER/Message/AnalysisRequest.cls       |  5 ++++
 src/ENCODER/WS/Service.cls                    |  1 +
 8 files changed, 44 insertions(+), 17 deletions(-)

diff --git a/encoder-ui/src/app/analyzer/analyzer.component.html b/encoder-ui/src/app/analyzer/analyzer.component.html
index e774093..dfb2cc1 100644
--- a/encoder-ui/src/app/analyzer/analyzer.component.html
+++ b/encoder-ui/src/app/analyzer/analyzer.component.html
@@ -4,7 +4,9 @@
             <form class="p-4" (ngSubmit)="onSubmit()" [formGroup]="textForm">
                 <div class="form-group">
                     <label for="textToAnalyze" class="font-weight-bold">{{'TEXT-ANALYZE' | transloco}}:</label>
-                    <textarea class="form-control" formControlName="TextToAnalyze" rows="8"></textarea>
+                    <textarea class="form-control" formControlName="TextToAnalyze" rows="8" name="textToAnalyze"></textarea>
+                    <label for="useLLM" class="font-weight-bold form-check-label">{{'USE-LLM' | transloco}}: </label>
+                    <input type="checkbox" class="btn btn-primary" class="form-check-input checkbox-form" name="useLLM" formControlName="UseLLM" value="true">
                 </div>
                 <div class="form-group pt-4 text-center">
                     <button type="submit" class="btn btn-primary" [ngClass]="{'disabled': loading}">
diff --git a/encoder-ui/src/app/analyzer/analyzer.component.scss b/encoder-ui/src/app/analyzer/analyzer.component.scss
index 017cd5f..28ecf3a 100644
--- a/encoder-ui/src/app/analyzer/analyzer.component.scss
+++ b/encoder-ui/src/app/analyzer/analyzer.component.scss
@@ -18,4 +18,8 @@
     max-height: 40vh;
     overflow-y: auto;
     border-radius: 8px;
+}
+
+.checkbox-form {
+    margin-left: .5em;
 }
\ No newline at end of file
diff --git a/encoder-ui/src/app/analyzer/analyzer.component.ts b/encoder-ui/src/app/analyzer/analyzer.component.ts
index 860fdad..898799a 100644
--- a/encoder-ui/src/app/analyzer/analyzer.component.ts
+++ b/encoder-ui/src/app/analyzer/analyzer.component.ts
@@ -41,30 +41,43 @@ export class AnalyzerComponent {
   }
 
   public textForm = new UntypedFormGroup({
-    TextToAnalyze: new UntypedFormControl('', {nonNullable: true, validators: [Validators.required]})
+    TextToAnalyze: new UntypedFormControl('', {nonNullable: true, validators: [Validators.required]}),
+    UseLLM: new UntypedFormControl('', {nonNullable: false, validators: [Validators.required]})
   })
 
   get textToAnalyze() {
     return this.textForm.get('TextToAnalyze');
   }
 
+  get useLLM() {
+    return this.textForm.get('UseLLM');
+  }
+
   onSubmit() {
     this.diagnostics = [];
     var textHTML = this.textToAnalyze?.value;
     var textOriginal = textHTML;
-    var textToProcess = this.textToAnalyze?.value.split(".").filter(Boolean);
+    var llmEnabled = this.useLLM?.value;
+    var textToProcess = [];
     var piecedTextToProcess: any[] = [];
-    
-    for (var index in textToProcess){
-      piecedTextToProcess = piecedTextToProcess.concat(textToProcess[index].split(","))
+
+    if (llmEnabled) {
+      piecedTextToProcess = [textOriginal]
     }
+    else {
+      textToProcess = this.textToAnalyze?.value.split(".").filter(Boolean);
+      for (var index in textToProcess){
+        piecedTextToProcess = piecedTextToProcess.concat(textToProcess[index].split(","))
+      }
+    }
+    
     var forReading = 100/(piecedTextToProcess.length);
     this.totalReceived = 0;
     this.error = false;
     this.loading = true;
     this.textAndDiagnosticList = [];
     const rawText = {
-      "Text": textOriginal
+      "Text": textOriginal,
     };
     this.irisService.saveRawText(rawText).subscribe({next: raw => {
       this.totalReceived = 0;
@@ -74,7 +87,8 @@ export class AnalyzerComponent {
           const textData = {
             "ID": raw.id,
             "Text": piecedTextToProcess[index],
-            "Language": this.translocoService.getActiveLang()
+            "Language": this.translocoService.getActiveLang(),
+            "UseLLM": llmEnabled
           };
           this.irisService.analyzeText(textData).subscribe({next: resp =>{
             this.totalReceived += forReading;
diff --git a/encoder-ui/src/assets/i18n/en.json b/encoder-ui/src/assets/i18n/en.json
index 199a70b..e71b963 100644
--- a/encoder-ui/src/assets/i18n/en.json
+++ b/encoder-ui/src/assets/i18n/en.json
@@ -38,5 +38,6 @@
     "IMPORTED-RECORDS": "Imported records",
     "VECTORIZED-RECORDS": "Vectorized records",
     "SIMILARITIES": "Similarities found for",
-    "MATCHES": "Matches"
+    "MATCHES": "Matches",
+    "USE-LLM": "Use LLM for diagnosis recognition"
 }
diff --git a/encoder-ui/src/assets/i18n/es.json b/encoder-ui/src/assets/i18n/es.json
index 3740670..3a965ab 100644
--- a/encoder-ui/src/assets/i18n/es.json
+++ b/encoder-ui/src/assets/i18n/es.json
@@ -38,5 +38,6 @@
     "IMPORTED-RECORDS": "Registros importados",
     "VECTORIZED-RECORDS": "Registros vectorizados",
     "SIMILARITIES": "Similitudes encontradas para",
-    "MATCHES": "Coincidencias"
+    "MATCHES": "Coincidencias",
+    "USE-LLM": "Usar LLM para el reconocimiento de diagnósticos"
 }
diff --git a/src/ENCODER/BP/AnalyzeTextProcess.cls b/src/ENCODER/BP/AnalyzeTextProcess.cls
index 19effb1..c0735bf 100644
--- a/src/ENCODER/BP/AnalyzeTextProcess.cls
+++ b/src/ENCODER/BP/AnalyzeTextProcess.cls
@@ -5,7 +5,12 @@ Method OnRequest(pInput As ENCODER.Message.AnalysisRequest, Output pOutput As EN
 {
         
         set pOutput = ##class(ENCODER.Message.AnalysisResponse).%New()
-        set result = ..AnalyzeTextSimplified(pInput.TextToAnalyze, pInput.AnalysisId, pInput.LanguageSelected)
+        if (pInput.UseLLM) {
+            set result = ..AnalyzeText(pInput.TextToAnalyze, pInput.AnalysisId, pInput.LanguageSelected)
+        }
+        else {
+            set result = ..AnalyzeTextSimplified(pInput.TextToAnalyze, pInput.AnalysisId, pInput.LanguageSelected)
+        }
         set pOutput.AnalysisRequestId = pInput.AnalysisId
         return $$$OK
 }
@@ -50,21 +55,15 @@ Method AnalyzeTextSimplified(text As %String, analysisId As %String, language As
     import requests
 
     try:
-        iris.cls("Ens.Util.Log").LogInfo("ENCODER.BP.AnalyzeTextProcess", "AnalyzeText", "Bottle neck 1")
         model = sentence_transformers.SentenceTransformer('/iris-shared/model/')
-        iris.cls("Ens.Util.Log").LogInfo("ENCODER.BP.AnalyzeTextProcess", "AnalyzeText", "Bottle neck 2")
         phrases = text.split(".")
         sqlsentence = ""
         # iris.cls("Ens.Util.Log").LogInfo("ENCODER.BP.AnalyzeTextProcess", "AnalyzeText", "Starting process")
         for phraseToAnalyze in phrases :
-            iris.cls("Ens.Util.Log").LogInfo("ENCODER.BP.AnalyzeTextProcess", "AnalyzeText", "Bottle neck 3")
             if phraseToAnalyze != "":
                 embedding = model.encode(phraseToAnalyze, normalize_embeddings=True).tolist()
-                iris.cls("Ens.Util.Log").LogInfo("ENCODER.BP.AnalyzeTextProcess", "AnalyzeText", "Bottle neck 4")
                 stmt = iris.sql.prepare("INSERT INTO ENCODER_Object.TextMatches (CodeId, Description, Similarity, AnalysisId, RawText) SELECT TOP 50 * FROM (SELECT CodeId, Description, VECTOR_DOT_PRODUCT(VectorDescription, TO_VECTOR(?, DECIMAL)) AS Similarity, ?, ? FROM ENCODER_Object.Codes) WHERE Similarity > 0.65 ORDER BY Similarity DESC")                    
-                iris.cls("Ens.Util.Log").LogInfo("ENCODER.BP.AnalyzeTextProcess", "AnalyzeText", "Bottle neck 5")
                 rs = stmt.execute(str(embedding), analysisId, phraseToAnalyze)        
-                iris.cls("Ens.Util.Log").LogInfo("ENCODER.BP.AnalyzeTextProcess", "AnalyzeText", "Bottle neck 6")
     except Exception as err:
         iris.cls("Ens.Util.Log").LogInfo("ENCODER.BP.AnalyzeTextProcess", "AnalyzeText", repr(err))
         return repr(err)
diff --git a/src/ENCODER/Message/AnalysisRequest.cls b/src/ENCODER/Message/AnalysisRequest.cls
index b024709..aca8fc0 100644
--- a/src/ENCODER/Message/AnalysisRequest.cls
+++ b/src/ENCODER/Message/AnalysisRequest.cls
@@ -7,6 +7,8 @@ Property AnalysisId As %String;
 
 Property LanguageSelected As %String;
 
+Property UseLLM As %Boolean;
+
 Storage Default
 {
 <Data name="AnalysisRequestDefaultData">
@@ -20,6 +22,9 @@ Storage Default
 <Value name="3">
 <Value>LanguageSelected</Value>
 </Value>
+<Value name="4">
+<Value>UseLLM</Value>
+</Value>
 </Data>
 <DefaultData>AnalysisRequestDefaultData</DefaultData>
 <Type>%Storage.Persistent</Type>
diff --git a/src/ENCODER/WS/Service.cls b/src/ENCODER/WS/Service.cls
index 3ebaa21..356ddab 100644
--- a/src/ENCODER/WS/Service.cls
+++ b/src/ENCODER/WS/Service.cls
@@ -152,6 +152,7 @@ ClassMethod AnalyzeText() As %Status
         set analysisRequest.TextToAnalyze = dynamicBody.%Get("Text")
         set analysisRequest.AnalysisId = dynamicBody.%Get("ID")
         set analysisRequest.LanguageSelected = dynamicBody.%Get("Language")
+        set analysisRequest.UseLLM = dynamicBody.%Get("UseLLM")
         
         // Creation of BS instance
         set status = ##class(Ens.Director).CreateBusinessService("ENCODER.BS.AnalyzeTextService", .instance)