Major change, added LLM with Ollama

docker-compose.yml

@@ -17,6 +17,8 @@ services:
     command: --check-caps false --ISCAgent false
     mem_limit: 30G
     memswap_limit: 32G
+    depends_on:
+      - ollama
 
   # web gateway container
   webgateway:
@@ -49,4 +51,15 @@ services:
       - 80:80
       - 443:443
     depends_on:
-      - iris
+      - iris
+
+  ## llm locally installed
+  ollama:
+    build:
+      context: .
+      dockerfile: ollama/Dockerfile
+    container_name: ollama
+    volumes:
+    - ./ollama/shared:/ollama-shared
+    ports:
+      - "11434:11434"

encoder-ui/src/app/analyzer/analyzer.component.ts

@@ -53,10 +53,8 @@ export class AnalyzerComponent {
     var textHTML = this.textToAnalyze?.value;
     var textOriginal = textHTML;
     var textToProcess = this.textToAnalyze?.value.split(".").filter(Boolean);
-    var piecedTextToProcess: any[] = [];
-    for (var index in textToProcess){
-      piecedTextToProcess = piecedTextToProcess.concat(textToProcess[index].split(","))
-    }
+    var piecedTextToProcess: any[] = [textOriginal];
+
     var forReading = 100/(piecedTextToProcess.length);
     this.totalReceived = 0;
     this.error = false;

iris/Dockerfile

@@ -6,7 +6,7 @@ USER root
 WORKDIR /opt/irisapp
 RUN chown -R irisowner:irisowner /opt/irisapp
 
-RUN apt-get update && apt-get install -y python3
+RUN apt-get update && apt-get install -y python3 && apt-get -y install curl
 
 # install required packages
 COPY --chown=$ISC_PACKAGE_MGRUSER:$ISC_PACKAGE_IRISGROUP /requirements.txt /

iris/shared/hl7/messagesa01_es.hl7

@@ -2,7 +2,7 @@ MSH|^~\&|HIS|HULP|EMPI||||ADT^A01|592956|P|2.5.1
 EVN|A01|
 PID|||1556655212^^^SERMAS^SN~922210^^^HULP^PI||GARCÍA PÉREZ^JUAN^^^||20150403|M|||PASEO PEDRO ÁLVAREZ 195 1 CENTRO^^LEGANÉS^MADRID^28379^SPAIN||555283055^PRN^^[email protected]|||||||||||||||||N|
 PV1||N
-DG1|1|||Hipertensión gestacional||A||
+DG1|1|||Tendinopatía del supraespinoso||A||
 
 
 MSH|^~\&|HIS|HULP|EMPI||||ADT^A01|628547|P|2.5.1

ollama/Dockerfile

@@ -0,0 +1,6 @@
+FROM ollama/ollama:latest
+
+COPY /ollama/entrypoint.sh /
+RUN chmod +x /entrypoint.sh
+
+ENTRYPOINT ["/entrypoint.sh"]

ollama/entrypoint.sh

@@ -0,0 +1,14 @@
+#!/bin/bash
+
+echo "Starting Ollama server..."
+ollama serve &
+SERVE_PID=$!
+
+echo "Waiting for Ollama server to be active..."
+while ! ollama list | grep -q 'NAME'; do
+  sleep 1
+done
+
+ollama pull llama3.2
+
+wait $SERVE_PID

requirements.txt

@@ -1,4 +1,4 @@
 sentence-transformers
 numpy
 pandas
-spacy
+spacy

src/ENCODER/BP/AnalyzeTextProcess.cls

@@ -15,46 +15,28 @@ Method AnalyzeText(text As %String, analysisId As %String, language As %String)
     import sentence_transformers
     import iris
     import spacy
+    import requests
 
     try:
-        lematize = ''
-        if language == 'es':
-            lematize = 'es_dep_news_trf'
-        else:
-            lematize = 'en_core_web_md'
+        url = "http://ollama:11434/api/generate"
+        data = {
+            "model": "llama3.2",
+            "prompt": "Extrae en formato CSV el texto literal de los diagnósticos encontrados en el paciente del siguiente texto sin mostrar diagnósticos que no se observen ni para los que no hay hallazgos patológicos, devuelve únicamente el CSV, evita fórmulas de cortesía: "+text,
+            "stream": False
+        }
+        response = requests.post(url, json=data)
+        analyzedText = response.json()
 
-        iris.cls("Ens.Util.Log").LogInfo("ENCODER.BP.AnalyzeTextProcess", "AnalyzeText", "Loading lematizer")
-        nlp = spacy.load(lematize)
-
         model = sentence_transformers.SentenceTransformer('/iris-shared/model/')
-        phrases = text.split(",")
+        phrases = analyzedText['response'].replace('"',"").split(",")
         sqlsentence = ""
         iris.cls("Ens.Util.Log").LogInfo("ENCODER.BP.AnalyzeTextProcess", "AnalyzeText", "Starting process")
         for phraseToAnalyze in phrases :
             if phraseToAnalyze != "":
-                doc = nlp(phraseToAnalyze)
-                phrase = ""
-                phrases = []
-                for token in doc:
-                    if (token.pos_ == "NOUN" or token.pos_ == "ADJ" or token.pos_ == "DET"):
-                        if phrase == "":
-                            phrase = token.text
-                        else :
-                            phrase += " "+token.text
-                    else :
-                        if phrase != "" and len(phrase.split(" ")) > 1:
-                            phrases.append(phrase)
-                        phrase = ""
-                embeddings = model.encode(phrases, normalize_embeddings=True)
-                embeddingList = embeddings.tolist()
-                i = 0
-                for text in phrases:
-                    iris.cls("Ens.Util.Log").LogInfo("ENCODER.BP.AnalyzeTextProcess", "AnalyzeText", text)
-                    sqlsentence = "INSERT INTO ENCODER_Object.TextMatches (CodeId, Description, Similarity, AnalysisId, RawText) SELECT TOP 50 * FROM (SELECT CodeId, Description, VECTOR_DOT_PRODUCT(VectorDescription, TO_VECTOR('"+str(embeddingList[i])+"', DECIMAL)) AS Similarity, '"+analysisId+"', '"+text+"' FROM ENCODER_Object.Codes) WHERE Similarity > 0.6 ORDER BY Similarity DESC"
-                    iris.cls("ENCODER.Utils.Manager").ExecuteInsertQuery(sqlsentence)
-                    iris.cls("Ens.Util.Log").LogInfo("ENCODER.BP.AnalyzeTextProcess", "AnalyzeText", "Sentence finished")
-
-                    i = i + 1
+                embedding = model.encode(phraseToAnalyze, normalize_embeddings=True).tolist()
+                sqlsentence = "INSERT INTO ENCODER_Object.TextMatches (CodeId, Description, Similarity, AnalysisId, RawText) SELECT TOP 50 * FROM (SELECT CodeId, Description, VECTOR_DOT_PRODUCT(VectorDescription, TO_VECTOR('"+str(embedding)+"', DECIMAL)) AS Similarity, '"+analysisId+"', '"+phraseToAnalyze+"' FROM ENCODER_Object.Codes) ORDER BY Similarity DESC"                    
+                iris.cls("ENCODER.Utils.Manager").ExecuteInsertQuery(sqlsentence)
+                iris.cls("Ens.Util.Log").LogInfo("ENCODER.BP.AnalyzeTextProcess", "AnalyzeText", sqlsentence)                    
     except Exception as err:
         iris.cls("Ens.Util.Log").LogInfo("ENCODER.BP.AnalyzeTextProcess", "AnalyzeText", repr(err))
         return repr(err)

src/ENCODER/Message/ConfigureRequest.cls

@@ -0,0 +1,4 @@
+Class ENCODER.Message.ConfigureRequest Extends Ens.Request
+{
+    Property Model As %String;
+}