Skip to content

Commit

Permalink
fixing output format
Browse files Browse the repository at this point in the history
  • Loading branch information
@pnrobinson
pnrobinson committed Aug 23, 2023
1 parent f9c44ba commit 89a7099
Show file tree
Hide file tree
Showing 3 changed files with 59 additions and 18 deletions.
2 changes: 2 additions & 0 deletions src/main/java/org/monarchinitiative/phenopacket2prompt/model/AdditionalConceptType.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -19,6 +19,7 @@ public enum AdditionalConceptType {
DIAGNOSTICS,
TREATMENT,
PMH,
FAMILY_HISTORY,
VERBATIM;


Expand All @@ -30,6 +31,7 @@ public static AdditionalConceptType of(String s) {
case "DIAGNOSTICS" -> DIAGNOSTICS;
case "TREATMENT" -> TREATMENT;
case "PMH" -> PMH;
case "FAMILY_HISTORY" -> FAMILY_HISTORY;
case "VERBATIM" -> VERBATIM;
default -> throw new PhenolRuntimeException("Unrecognised concept \"" + concept + "\"");
};
Expand Down
16 changes: 16 additions & 0 deletions ...va/org/monarchinitiative/phenopacket2prompt/querygen/qfactory/AbstractQueryGenerator.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -89,6 +89,16 @@ protected String getPersonIntroduction() {
return String.format("%s presented with the following signs and symptoms:\n", person_string);
}

private String stripFamilyHistory(String originalSeg) {
List<String> validLines = new ArrayList<>(); // everything but family history
String [] lines = originalSeg.split("\\.");
for (var line : lines) {
if (line.toLowerCase().contains("family history")) continue;
validLines.add(line);
}
return String.join(". ", validLines);
}

protected Map<String, String> timeSegments(String vignette, List<TimePoint> timePointList) {
Map<String, String> timeSegments = new LinkedHashMap<>(); // ordered map
String nextStart = "";
Expand All @@ -97,6 +107,7 @@ protected Map<String, String> timeSegments(String vignette, List<TimePoint> time
int s = timePoint.start();
int e = timePoint.end();
String seg = nextStart + vignette.substring(lastEnd, s);
seg = stripFamilyHistory(seg);
lastEnd = e + 1;
timeSegments.put(nextStart, seg.strip());
nextStart = timePoint.point();
Expand All @@ -113,6 +124,11 @@ protected Map<String, String> timeSegments(String vignette, List<TimePoint> time
* @return A string formatted as X, Y, and Z.
*/
protected String getOxfordCommaList(Set<String> items) {
if (items.size() == 2) {
// no comma if we just have two items.
// one item will work with the below code
return String.join(" and ", items) + ".";
}
StringBuilder sb = new StringBuilder();
String symList = String.join(", ", items);
int jj = symList.lastIndexOf(", ");
Expand Down
59 changes: 41 additions & 18 deletions ...a/org/monarchinitiative/phenopacket2prompt/querygen/qfactory/TextPlusManualGenerator.java
View file
Original file line number Diff line number Diff line change
@@ -1,9 +1,11 @@
package org.monarchinitiative.phenopacket2prompt.querygen.qfactory;

import org.checkerframework.checker.units.qual.A;
import org.monarchinitiative.fenominal.core.TermMiner;
import org.monarchinitiative.phenol.base.PhenolRuntimeException;
import org.monarchinitiative.phenol.ontology.data.Ontology;
import org.monarchinitiative.phenopacket2prompt.model.AdditionalConceptI;
import org.monarchinitiative.phenopacket2prompt.model.AdditionalConceptType;
import org.monarchinitiative.phenopacket2prompt.nejm.NejmCaseReportFromPdfFilterer;
import org.monarchinitiative.phenopacket2prompt.querygen.TimePoint;
import org.monarchinitiative.phenopacket2prompt.querygen.TimePointParser;
Expand All @@ -12,10 +14,7 @@
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

import java.util.HashSet;
import java.util.List;
import java.util.Map;
import java.util.Set;
import java.util.*;
import java.util.function.Predicate;
import java.util.stream.Collectors;

Expand All @@ -24,10 +23,20 @@ public class TextPlusManualGenerator extends AbstractQueryGenerator {
private final String promptText;

private final Set<AdditionalConceptI> additionalConcepts;
private final Set<String> pmh;
private final Set<String> familyHistory;

private final List<String> outputLines;


public TextPlusManualGenerator(NejmCaseReportFromPdfFilterer filterer, String id, TermMiner miner, Ontology hpo) {
super(filterer, id, miner, hpo);
this.outputLines = new ArrayList<>();
this.pmh = new HashSet<>();
familyHistory = filterer.getAdditionalConcepts().stream()
.filter(a -> a.conceptType() == AdditionalConceptType.FAMILY_HISTORY)
.map(AdditionalConceptI::insertText)
.collect(Collectors.toSet());
this.additionalConcepts = filterer.getAdditionalConcepts();
String phenotext = getPhenopacketTextWithAdditions();
promptText = String.format("%s%s", QUERY_HEADER, phenotext);
Expand All @@ -47,8 +56,7 @@ protected String getPhenopacketTextWithAdditions() {
vignette = vignette.substring(ii + 1);
List<TimePoint> timePointList = timePointParser.getTimePoints(vignette);

StringBuilder sb = new StringBuilder();
sb.append(firstSentence).append("\n");


try {
//Map<String, String> timeSegments = timeSegments(starts, ends, vignette, start2pointMap);
Expand All @@ -62,13 +70,29 @@ protected String getPhenopacketTextWithAdditions() {
if (description.length() > MIN_DESCRIPTION_LENGTH) {
String output = getPhenopacketBasedQuerySegmentWithAdditions(timePoint, description);
if (output.isEmpty()) continue;
sb.append(output).append("\n");
//sb.append(output).append("\n");
outputLines.add(output);
}
}
} catch (Exception eee) {
System.out.printf("[ERROR(TextPlusManualGenerator.java] Could not parse time segments for because of %s", eee.getMessage());
System.exit(1);
}
StringBuilder sb = new StringBuilder();
sb.append(firstSentence).append("\n");
if (pmh.size() > 0) {
sb.append("The past medical history was notable for ")
.append(getOxfordCommaList(pmh))
.append("\n");
}
if (familyHistory.size() > 0) {
for (String item: familyHistory) {
sb.append(item).append("\n");
}
}
for (var line : outputLines) {
sb.append(line);
}
return sb.toString();
}

Expand All @@ -82,7 +106,6 @@ protected String getPhenopacketBasedQuerySegmentWithAdditions(String presentatio
Set<String> treatment = new HashSet<>();
Set<String> verbatim = new HashSet<>();
/* Past medical history */
Set<String> pmh = new HashSet<>();

Set<String> observed_terms = pfeatures.stream()
.filter(Predicate.not(PhenotypicFeature::getExcluded))
Expand All @@ -95,25 +118,24 @@ protected String getPhenopacketBasedQuerySegmentWithAdditions(String presentatio
.map(OntologyClass::getLabel)
.collect(Collectors.toSet());
for (var addcon : this.additionalConcepts ) {
LOGGER.error("TOP {}", addcon);
if (input.contains(addcon.originalText())) {
LOGGER.error("FOUND INPUT {}", addcon);
switch (addcon.conceptType()) {
case PHENOTYPE -> observed_terms.add(addcon.insertText());
case EXCLUDE -> excluded_terms.add(addcon.insertText());
case DIAGNOSTICS -> diagnostics.add(addcon.insertText());
case TREATMENT -> treatment.add(addcon.insertText());
case VERBATIM -> verbatim.add(addcon.insertText());
case PMH -> pmh.add(addcon.insertText());
case PMH -> {
// do not repeat the PMH even if the original text mentions it more than once
if (!pmh.contains(addcon.originalText())) {
pmh.add(addcon.insertText());
}
}
case FAMILY_HISTORY -> familyHistory.add(addcon.insertText());
}
}
}
StringBuilder sb = new StringBuilder();
if (! pmh.isEmpty()) {
sb.append("The past medical history was notable for ");
sb.append(getOxfordCommaList(pmh));
sb.append(".\n");
}
String capitalizedTimepoint;
if (presentationTimeDescription.equalsIgnoreCase("Examination was notable for")) {
presentationTimeDescription = "On examination";
Expand Down Expand Up @@ -146,15 +168,16 @@ protected String getPhenopacketBasedQuerySegmentWithAdditions(String presentatio
} else {
sb.append("The following signs and symptoms were excluded: ");
}
sb.append(excludededSymptoms).append(" ");
sb.append(excludededSymptoms).append("\n");
}
if (! diagnostics.isEmpty()) {
sb.append("The following diagnostic observations were made: \n");
sb.append(getOxfordCommaList(diagnostics));
}
if (! treatment.isEmpty()) {
sb.append("The following treatments were administered: \n");
sb.append("The following treatments were administered: ");
sb.append(getOxfordCommaList(treatment));
sb.append("\n");
}
if (! verbatim.isEmpty()) {
for (var v : verbatim)
Expand Down

0 comments on commit 89a7099

Please sign in to comment.