Repository Methods Update¶

Last Updated: 2026-01-19
Status: ✅ COMPLETED

Overview¶

Added findAll() methods to repositories to support database existence checks in the synthetic data generator optimization.

Changes Made¶

1. ICD10CodeRepository¶

File: src/main/java/com/berdachuk/medexpertmatch/medicalcoding/repository/ICD10CodeRepository.java

Added Method:

/**
 * Finds all ICD-10 codes.
 *
 * @return List of all ICD-10 codes
 */
List<ICD10Code> findAll();

Implementation: src/main/java/com/berdachuk/medexpertmatch/medicalcoding/repository/impl/ICD10CodeRepositoryImpl.java

@Override
public List<ICD10Code> findAll() {
    return namedJdbcTemplate.query(findAllSql, Map.of(), icd10CodeMapper);
}

SQL File: src/main/resources/sql/medicalcoding/findAll.sql

SELECT id, code, description, category, parent_code, related_codes
FROM medexpertmatch.icd10_codes
ORDER BY code

2. MedicalSpecialtyRepository¶

Status: ✅ Already existed - no changes needed

File: src/main/java/com/berdachuk/medexpertmatch/doctor/repository/MedicalSpecialtyRepository.java

Existing Method:

/**
 * Finds all medical specialties.
 *
 * @return List of all medical specialties
 */
List<MedicalSpecialty> findAll();

Implementation: src/main/java/com/berdachuk/medexpertmatch/doctor/repository/impl/MedicalSpecialtyRepositoryImpl.java

@Override
public List<MedicalSpecialty> findAll() {
    return namedJdbcTemplate.query(findAllSql, medicalSpecialtyMapper);
}

SQL File: src/main/resources/sql/doctor/medicalspecialty/findAll.sql

SELECT id, name, normalized_name, description, icd10_code_ranges, related_specialties
FROM medexpertmatch.medical_specialties
ORDER BY name

Usage in SyntheticDataGenerator¶

Both findAll() methods are used in the synthetic data generator to check if sufficient data already exists in the database before making LLM calls:

ICD-10 Codes Check¶

// In generateIcd10Codes()
int existingCount = icd10CodeRepository.findAll().size();
int targetCount = calculateTargetIcd10Count(size);

if(existingCount >=targetCount){
        log.

info("Sufficient ICD-10 codes exist in database ({} >= {}), loading existing codes and skipping LLM generation",
     existingCount, targetCount);

List<ICD10Code> existingCodes = icd10CodeRepository.findAll();
extendedIcd10Codes =existingCodes.

stream()
            .

map(ICD10Code::code)
            .

distinct()
            .

limit(targetCount)
            .

collect(Collectors.toList());
        return;
        }

Medical Specialties Check¶

// In generateMedicalSpecialties()
int existingCount = medicalSpecialtyRepository.findAll().size();
int targetCount = calculateTargetSpecialtyCount(size);

if(existingCount >=targetCount){
        log.

info("Sufficient medical specialties exist in database ({} >= {}), skipping LLM generation",
     existingCount, targetCount);
    return;
            }

Benefits¶

Reduced LLM Calls: Eliminates unnecessary LLM API calls when data already exists
Database Reuse: Leverages existing data instead of regenerating
Performance: Faster generation when data is already available
Cost Savings: Reduces API costs for repeated generations

Testing¶

Both methods are tested through:

Integration tests in SyntheticDataGeneratorIT
Database existence checks in generateIcd10Codes() and generateMedicalSpecialties()