Conformance Verification of Healthcare Application Code Against Clinical Guidelines: A Case Study of Blood Pressure Classification
- 1Department of Information Technology and Health Informatics, Allied Health Sciences, Federal University of Medicine and Medical Sciences, Abeokuta, Nigeria.
Abstract
Healthcare applications increasingly implement clinical guidelines as software-based decision rules. However, validation of measurement accuracy does not establish whether the implemented classification logic conforms to the corresponding clinical guideline. This study evaluated whether exhaustive conformance verification can identify discrepancies between healthcare application code and a named clinical specification. The 2023 European Society of Hypertension (ESH) guideline was formalized as a deterministic reference oracle for blood pressure classification. An exhaustive integer input space comprising systolic blood pressure values from 60 to 260 mmHg and diastolic blood pressure values from 40 to 160 mmHg was generated, yielding 24,321 ordered input pairs. Fifty-seven open-source blood pressure classification applications were evaluated against the reference oracle. Binary hypertensive status and, where applicable, multi-category classification were assessed independently. Disagreements were traced to the corresponding decision branches and characterized according to their structural pattern and clinical direction. All 57 applications reproduced the binary ESH hypertensive/non-hypertensive decision across the complete input space, resulting in 100.000% sensitivity, specificity, and accuracy. Category-level analysis was possible for 41 applications. Fifteen applications achieved ≥99.9% conformance, 10 achieved 80–99.9% conformance, and 16 achieved <80% conformance. The lowest category-level conformance was 4.354%. The identified discrepancies primarily involved under-classification and were associated with cascade-ordering errors, coverage gaps, and unreachable crisis branches. One implementation had 23,190 of 24,321 input combinations classified outside the applicable ESH categories. Binary agreement with a clinical guideline did not establish complete conformance at the category level. Exhaustive verification provided a deterministic and reproducible method for evaluating bounded clinical classification logic and for tracing discrepancies to specific implementation structures. This approach may complement measurement validation, usability assessment, security testing, and other forms of healthcare software assurance.
Related
- Adherence to HbA1c testing guidelines in type 2 diabetes patients: A mixed-methods retrospective cross-sectional study in Bahrain’s primary healthcare centers
- Knowledge, Awareness, and Perception toward Ethical Guidelines for Artificial Intelligence in Biomedical Research and Healthcare among Dental Faculty and Postgraduates – A Cross-sectional Study
- Hypertension disease natural history and healthcare utilization when treatment guidelines do not account for race
- Application of relative energy deficiency in sport (REDs) guidelines: a scoping review
- Optimising summary infographics for clinical practice guidelines on child health: qualitative user testing study with healthcare workers in Malawi, Nigeria and South Africa
BibTeX
@article{ojediran2026conformance,
title = {{Conformance Verification of Healthcare Application Code Against Clinical Guidelines: A Case Study of Blood Pressure Classification}},
author = {Alaba Ojediran},
journal = {InfoScience Trends},
year = {2026},
month = sep,
doi = {10.66224/ist.202603.14.03},
url = {https://doi.org/10.66224/ist.202603.14.03},
note = {CC BY},
}
RIS
TY - JOUR
TI - Conformance Verification of Healthcare Application Code Against Clinical Guidelines: A Case Study of Blood Pressure Classification
AU - Alaba Ojediran
PY - 2026
DA - 2026/09/01
JO - InfoScience Trends
DO - 10.66224/ist.202603.14.03
UR - https://doi.org/10.66224/ist.202603.14.03
AB - Healthcare applications increasingly implement clinical guidelines as software-based decision rules. However, validation of measurement accuracy does not establish whether the implemented classification logic conforms to the corresponding clinical guideline. This study evaluated whether exhaustive conformance verification can identify discrepancies between healthcare application code and a named clinical specification. The 2023 European Society of Hypertension (ESH) guideline was formalized as a deterministic reference oracle for blood pressure classification. An exhaustive integer input space comprising systolic blood pressure values from 60 to 260 mmHg and diastolic blood pressure values from 40 to 160 mmHg was generated, yielding 24,321 ordered input pairs. Fifty-seven open-source blood pressure classification applications were evaluated against the reference oracle. Binary hypertensive status and, where applicable, multi-category classification were assessed independently. Disagreements were traced to the corresponding decision branches and characterized according to their structural pattern and clinical direction. All 57 applications reproduced the binary ESH hypertensive/non-hypertensive decision across the complete input space, resulting in 100.000% sensitivity, specificity, and accuracy. Category-level analysis was possible for 41 applications. Fifteen applications achieved ≥99.9% conformance, 10 achieved 80–99.9% conformance, and 16 achieved <80% conformance. The lowest category-level conformance was 4.354%. The identified discrepancies primarily involved under-classification and were associated with cascade-ordering errors, coverage gaps, and unreachable crisis branches. One implementation had 23,190 of 24,321 input combinations classified outside the applicable ESH categories. Binary agreement with a clinical guideline did not establish complete conformance at the category level. Exhaustive verification provided a deterministic and reproducible method for evaluating bounded clinical classification logic and for tracing discrepancies to specific implementation structures. This approach may complement measurement validation, usability assessment, security testing, and other forms of healthcare software assurance.
SN - 3041-9875
ER -