**9. Conclusions**

The proposed approach using advanced parallel *k*-means clustering for pre-processing medical data, combined with logistic regression and YOLOv4 for classification and image detection, respectively, effectively improved the performance of these algorithms, particularly when applied to large medical datasets. The results of the classification task showed that the approach was able to accurately classify the medical data, and the results of the image detection

task using X-ray and CT scan images showed that the approach was able to effectively detect and classify the medical images. The use of advanced parallel *k*-means pre-processing and acceleration of the neural engine processor contributed to the improved accuracy and efficiency of the approach. This approach has the potential to significantly impact the field of healthcare, as it can aid in diagnostics, treatment planning, and disease monitoring. Further research and evaluation on larger and more diverse medical datasets could reveal additional benefits and potential applications. While the proposed solution has shown promise in improving the accuracy and efficiency of these tasks on large medical datasets, there were still limitations that should be considered. One limitation was the hardware dependency, as the acceleration of the *k*-means clustering was highly dependent on the neural engine processor, multi-core processor, and the operating system's support for hardware management. Another limitation was the ability to improve 24-bit color images, which require a different number of *k*-values and could affect the clustering performance negatively.

**Author Contributions:** Conceptualization, F.H.A., M.M.H. and L.A.; methodology, F.H.A., M.M.H. and L.A.; software, F.H.A., M.M.H. and L.A.; validation, F.H.A., M.M.H. and L.A.; formal analysis, F.H.A., M.M.H. and L.A.; investigation, F.H.A., M.M.H. and L.A.; resources, F.H.A., M.M.H. and L.A.; data curation, F.H.A., M.M.H. and L.A.; writing—original draft preparation, F.H.A., M.M.H. and L.A.; writing—review and editing, F.H.A., M.M.H. and L.A.; visualization, F.H.A., M.M.H. and L.A.; supervision M.M.H.; project administration, F.H.A., M.M.H. and L.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** Laith Alzubaidi would like to acknowledge the support received through the following funding schemes of Australian Government: Australian Research Council (ARC) Industrial Transformation Training Centre (ITTC) for Joint Biomechanics under grant IC190100020 and QUT ECR SCHEME 2022, The Queensland University of Technology.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** COVID-19 Research Challenge: (https://www.kaggle.com/datasets/ allen-institute-for-ai/CORD-19-research-challenge); Large COVID- 19 CT (https://www.kaggle. com/datasets/maedemaftouni/large-covid19-ct-slice-dataset); COVIDx- CT (https://www.kaggle. com/datasets/hgunraj/covidxct); CT- LOW- Dose (https://www.kaggle.com/datasets/andrewmvd/ ct-low-dose-reconstruction); Google Play Store (https://www.kaggle.com/datasets/gauthamp10/ google-playstore-apps); KDD99 (https://datahub.io/machine-learning/kddcup99).

**Conflicts of Interest:** The authors declare no conflict of interest.
