Prediction of Early Mortality in Patients Undergoing Right Hemicolectomy Using Machine Learning Algorithms

doi:10.21203/rs.3.rs-4722591/v1

Download PDF

Short Report

Prediction of Early Mortality in Patients Undergoing Right Hemicolectomy Using Machine Learning Algorithms

https://doi.org/10.21203/rs.3.rs-4722591/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Aim:This study aims to determine whether early mortality in patients undergoing right hemicolectomy can be predicted using artificial intelligence (machine learning) algorithms.

Method:The study included all cases of right hemicolectomy or extended right hemicolectomy performed in our clinic between January 2019 and December 2023. Data were collected retrospectively from a prospectively maintained database. Patients undergoing surgeries other than right hemicolectomy were excluded. A database was created using basic clinical data and processed in the Google Colab environment using TensorFlow, Keras, Pandas, Numpy, and Scikit-learn libraries. The TensorFlow. Keras Sequential model was used with “Dense” layers, and the “Adam” optimizer was chosen for optimization. Eighty percent of the database was used for training, and 20% was used for testing.

Results:The database included 410 patients, of whom 258 (62.9%) were male, and 152 (37.1%) were female, with a mean age of 63.5±14.2 years. Early mortality occurred in 39 (9.5%) patients. Data from 307 patients were used for machine learning training, and data from 103 patients were used for testing. The machine-learning model predicted early mortality with an accuracy range of 91% to 95% using basic clinical parameters..

Conclusion:This study demonstrates that early mortality in patients undergoing right hemicolectomy can be predicted with high accuracy using machine learning algorithms. The above 90% accuracy rate achieved using basic clinical parameters indicates the potential of this algorithm as a supportive tool in clinical decision-making processes.

Right Hemicolectomy

Colon Cancer

Inflammatory Bowel Diseases

Machine Learning

Artificial Intelligence

Right hemicolectomy, a surgical procedure that involves removing the right side of the colon, is commonly performed to treat various conditions such as colon cancer, Crohn's disease, and other localized colonic pathologies. Despite being a routine operation, it carries significant risks, including postoperative complications and mortality. The ability to predict early mortality following right hemicolectomy can significantly improve patient outcomes by allowing clinicians to identify high-risk patients and implement targeted interventions.

Traditional risk assessment models, such as the American Society of Anesthesiologists (ASA) score and the Charlson Comorbidity Index, provide valuable insights but are often based on static parameters and may not fully capture the complexity of patient-specific factors and intraoperative variables. These traditional models may also be limited by subjective assessments and variability between practitioners. Recent advances in artificial intelligence, particularly machine learning (ML), offer new avenues for enhancing predictive accuracy by leveraging large datasets and identifying complex patterns that may be overlooked by traditional methods (1–5).

Machine learning, a subset of artificial intelligence, enables computers to learn from and make predictions based on data. By using algorithms that iteratively learn from data, machine learning models can improve their accuracy over time. These models are particularly useful in medical research, where they can analyse vast amounts of patient data to identify subtle correlations and predictors of outcomes (6–15). In the context of right hemicolectomy, machine learning can be used to develop predictive models that incorporate a wide range of clinical variables, thereby providing a more comprehensive risk assessment tool.

This study aims to explore the feasibility and accuracy of using machine learning algorithms to predict early mortality in patients undergoing right hemicolectomy. By retrospectively analysing data from a prospectively maintained database, we seek to determine whether machine learning can provide a reliable and practical tool for risk stratification in surgical patients. Our approach leverages commonly available clinical parameters to develop a model that can be easily integrated into clinical workflows, potentially enhancing the decision-making process and improving patient outcomes.

Study Design and Population

This retrospective study included all patients who underwent right hemicolectomy or extended right hemicolectomy in our clinic between January 2019 and December 2023. Patients who underwent additional surgical procedures were excluded to ensure a homogenous study population.

Data Collection

Clinical data were prospectively maintained in a database and included patient demographics, comorbidities, preoperative laboratory values, and perioperative details. Data extraction and analysis were conducted retrospectively. Early mortality was defined as deaths within the postoperative 30 days.

Institutional ethics committee approved the study design(101 − 24).Since the study was retrospective, no additional intervention was performed other than routine patient care. Informed consent is not applicable because information is anonymized and the submission does not include images that may identify the person.

Machine Learning Model

The dataset was processed using Google Colab with TensorFlow, Keras, Pandas, Numpy, and Scikit-learn libraries (Fig. 1). A Sequential model from TensorFlow. Keras was used, incorporating Dense layers to build the neural network. The “Adam” optimizer was selected for its efficiency in handling sparse gradients (Fig. 2). The dataset was split into training (80%) and testing (20%) subsets to evaluate model performance.

Demographics and Clinical Characteristics

During the study period, 419 patients underwent right hemicolectomy in our clinic; nine of these patients were excluded from the study due to exclusion criteria. Therefore, the study involved 410 patients with an average age of 63.54 years (± 14.19). Among them, 258 (62.9%) were male, while 152 (37.1%) were female. The distribution of procedures in terms of emergency was as follows: 151 (36.8%) were emergent and 259 (63.2%) were elective. The majority of surgeries were performed using an open technique (91.9%, n = 377), while laparoscopy accounted for 8.1% (n = 33) of the procedures.

Co-Morbidities

Hypertension (HT) was present in 141(34.4%) of patients. Sixty-two (15.1%) had a history of coronary artery disease (CAD). Diabetes Mellitus (DM) was diagnosed in 95 (23.2%) of patients. Chronic Obstructive Pulmonary Disease (COPD) was observed in 32 (7.8%) of cases. Chronic Renal Failure (CRF) was less prevalent, affecting 9 (%2.2) of patients.

Surgical Indications and Outcomes

Malignancy was a common indication for surgery, present in 82.9% (n = 340) of cases. Inflammatory Bowel Disease (IBD) was identified in 4.9% (n = 20) of patients. Postoperative complications included the formation of an ostomy (6.3%, n = 26) and surgical site infections (6.3%, n = 26). Anastomosis leak occurred in 2.9% (n = 12) of cases. All baseline demographic and clinical characteristics are listed in Table 1.

Table 1

Demographics, clinical characteristics, and outcomes of surgical procedures
Parameters			Values
Age (Years ± SD)			63.54	± 14.19
Gender		M	258	62.9%
Gender		F	152	37.1%
Emergency		Emergent	151	36.8%
Emergency		Elective	259	63.2%
Tecnique		Laparoscopy	33	8,1%
Tecnique		Open	377	91,9%
Co-Morbidities	HT	Yes	141	34.4%
	HT	No	269	65.6%
	CAD	Yes	62	15.1%
	CAD	No	348	84.9%
	DM	Yes	95	23.2%
	DM	No	315	76.8%
	COPD	Yes	32	7.8%
	COPD	No	378	92.2%
	CRF	Yes	9	2.2%
	CRF	No	401	97.8%
Malignancy		Yes	340	82.9%
Malignancy		No	70	17.1%
IBD		Yes	20	4.9%
IBD		No	390	95.1%
Ostomy		Yes	26	6.3%
Ostomy		No	384	93.7%
SSI		Yes	26	6.3%
SSI		No	384	93.7%
Anastomosis_Leak		Yes	12	2.9%
Anastomosis_Leak		No	398	97.1%
Hospital Stay (Days ± SD)			8.6±	± 7
ICU		Yes	146	35.6%
ICU		No	264	64.4%
Early Mortality		Yes	39	9.5%
Early Mortality		No	371	90.5%

Abbreviations: HT (Hypertension), CAD (Coronary Artery Disease), DM (Diabetes Mellitus), COPD (Chronic Obstructive Pulmonary Disease), CRF (Chronic Renal Failure), IBD (Inflammatory Bowel Disease), SSI (Surgical Site Infection), ICU (Intensive Care Unit). Early mortality refers to mortality within the postoperative 30 days.

Model Performance

The machine learning model was trained on data from 307 patients and tested on 103 patients. The model achieved an accuracy range of 91–95% in predicting early mortality using basic clinical parameters, demonstrating its potential effectiveness in a clinical setting (Figs. 2–4).

The utilization of artificial intelligence (AI) in healthcare has revolutionized the way medical professionals diagnose, treat, and manage diseases (1–15). AI algorithms can analyse vast amounts of medical data with remarkable speed and accuracy, aiding in early detection of illnesses and predicting patient outcomes. From image recognition in radiology to natural language processing in electronic health records, AI technologies streamline processes, reduce errors, and improve patient care (10). Moreover, AI-powered systems can personalize treatment plans based on individual patient characteristics, leading to more effective and efficient healthcare delivery. As AI continues to advance, its integration into healthcare promises even greater advancements in disease prevention, diagnosis, and treatment, ultimately enhancing the overall quality of patient care.

The findings of this study indicate that machine learning algorithms can accurately predict early mortality in patients undergoing right hemicolectomy using basic clinical parameters. The model achieved a 95% accuracy rate, demonstrating its potential effectiveness as a supportive tool in clinical decision-making. These results are promising and suggest that machine learning can provide a valuable adjunct to traditional risk assessment methods.

Validation and Generalizability

The high accuracy of the model in our dataset is encouraging, but further validation is required to ensure its reliability and applicability across different clinical settings. External validation studies involving larger and more diverse patient populations are essential to confirm the model's robustness. Additionally, the model should be tested in various healthcare environments, including community hospitals and tertiary care centres, to ensure its generalizability and adaptability to different clinical workflows.

Ethical and Legal Considerations

The integration of AI-based tools in healthcare raises several ethical and legal issues that must be addressed. Patient consent and data privacy are paramount concerns, particularly given the sensitive nature of medical data. Ensuring that patient data is anonymized and securely stored is critical to maintaining confidentiality. Moreover, the transparency and interpretability of the model's predictions are crucial for building trust among clinicians and patients. It is essential that healthcare providers understand how the model makes its predictions and can explain these predictions to patients. Regulatory frameworks must also be developed to govern the use of AI in healthcare, ensuring that these technologies are used responsibly and ethically.

Clinical Implications

The ability to accurately predict early mortality can have significant clinical implications. High-risk patients identified by the model could benefit from enhanced perioperative care, including closer monitoring, more aggressive management of comorbidities, and tailored postoperative interventions. This proactive approach could potentially reduce complication rates and improve overall outcomes. Furthermore, the model could assist in shared decision-making processes by providing patients and their families with more accurate information about surgical risks, thereby facilitating informed consent.

Future Directions

Future research should focus on refining the model by incorporating additional clinical variables, such as genetic markers, intraoperative factors, and postoperative recovery metrics. Combining machine learning with other advanced technologies, such as natural language processing and electronic health record integration, could further enhance predictive accuracy and clinical utility. Collaborations between data scientists, clinicians, and ethicists will be crucial in addressing the multifaceted challenges associated with implementing AI in healthcare.

Limitations

This study has several limitations that should be acknowledged. The retrospective nature of the data collection may introduce biases related to data completeness and accuracy. Additionally, the model's performance may be influenced by the specific characteristics of the patient population in our clinic, potentially limiting its applicability to other settings. Prospective studies are needed to validate the model's predictive capabilities in real-time clinical practice.

This study demonstrates the potential of machine learning algorithms to predict early mortality in patients undergoing right hemicolectomy with high accuracy using basic clinical parameters. The promising results highlight the potential of these algorithms as supportive tools in clinical decision-making, providing a means to enhance patient care and optimize surgical outcomes. However, extensive validation and careful consideration of ethical, legal, and practical issues are necessary before these tools can be widely adopted in clinical practice. By addressing these challenges, we can harness the power of AI to improve patient outcomes and advance the field of surgical risk assessment.

Data availability

The datasets analysed during the current study are available from the corresponding author on reasonable request.

Author Contributions:

a. Substantial contributions to conception and design, or acquisition of data, or analysis and interpretation of data: M.K, M.A, D.I, H.U

b. Drafting the article or revising it critically for important intellectual content: M.K, H.F.K, H.U

c. Final approval of the version to be published: H.F.K

Competing interests

The authors declare no competing interests.

Conflict of interest

The authors declare that they did not receive any financial and non-financial interest.

Ethical approval

Institutional ethics committee approved the study design (101-24). Since the study was retrospective, no additional intervention was performed other than routine patient care.

Human ethics and consent to participate declaration

Not applicable.

Lo Vercio, L., Amador, K., Bannister, J. J., et al. (2020). Supervised machine learning tools: A tutorial for clinicians. Journal of Neural Engineering, 19(6), 17. https://doi.org/10.1088/1741-2552/abbff2.
Larvie, M. (2019). Machine learning in radiology: Resistance is futile. Radiology, 290(2), 465-466. https://doi.org/10.1148/radiol.2018182312. Epub 2018 Nov 6. PMID: 30398438.
An, Q., Rahman, S., Zhou, J., & Kang, J. J. (2023). A comprehensive review on machine learning in healthcare industry: Classification, restrictions, opportunities and challenges. Sensors (Basel), 23(9), 4178. https://doi.org/10.3390/s23094178.
Kang, J., Schwartz, R., Flickinger, J., & Beriwal, S. (2015). Machine learning approaches for predicting radiation therapy outcomes: A clinician's perspective. International Journal of Radiation Oncology, Biology, Physics, 93(5), 1127-1135. https://doi.org/10.1016/j.ijrobp.2015.07.2286. Epub 2015 Nov 11.
Saillard, C., Schmauch, B., Laifa, O., et al. (2020). Predicting survival after hepatocellular carcinoma resection using deep learning on histological slides. Hepatology, 72(6), 2000-2013. https://doi.org/10.1002/hep.31207
Silva, G. F. S., Fagundes, T. P., Teixeira, B. C., & Chiavegatto Filho, A. D. P. (2022). Machine learning for hypertension prediction: A systematic review. Current Hypertension Reports, 24(11), 523-533. https://doi.org/10.1007/s11906-022-01212-6
Deepa, C., & Gunavatti, A. (2022). Systematic review on machine learning and deep learning techniques in cancer survival prediction. Progress in Biophysics and Molecular Biology, 174, 62-71. https://doi.org/10.1016/j.pbiomolbio.2022.07.004
Chang, C. C., & Chen, Y. C. (2020). Advanced machine learning in prediction of second primary cancer in colorectal cancer. Studies in Health Technology and Informatics, 270, 1191-1192. https://doi.org/10.3233/SHTI200357
Mohsen, F., Biswas, M. R., Ali, H., Alam, T., Househ, M., & Shah, Z. (2022). Customized and automated machine learning-based models for diabetes type 2 classification. Studies in Health Technology and Informatics, 295, 517-520. https://doi.org/10.3233/SHTI220779
Chen, K. A., Joisa, C. U., Stem, J. M., Guillem, J. G., Gomez, S. M., & Kapadia, M. R. (2023). Improved prediction of surgical-site infection after colorectal surgery using machine learning. Diseases of the Colon & Rectum, 66(3), 458-466. https://doi.org/10.1097/DCR.0000000000002559
Thompson, H. M., Kim, J. K., Jimenez-Rodriguez, R. M., Garcia-Aguilar, J., & Veeraraghavan, H. (2023). Deep learning-based model for identifying tumors in endoscopic images from patients with locally advanced rectal cancer treated with total neoadjuvant therapy. Diseases of the Colon & Rectum, 66(3), 383-391. https://doi.org/10.1097/DCR.0000000000002295. PMCID: PMC10185333.
Kayikcioglu, E., Onder, A. H., Bacak, B., & Serel, T. A. (2024). Machine learning for predicting colon cancer recurrence. Surgical Oncology, 54, 102079. https://doi.org/10.1016/j.suronc.2024.102079
Pacal, I., Karaboga, D., Basturk, A., Akay, B., & Nalbantoglu, U. (2020). A comprehensive review of deep learning in colon cancer. Computers in Biology and Medicine, 126, 104003. https://doi.org/10.1016/j.compbiomed.2020.104003
Wong, D., & Yip, S. (2018). Machine learning classifies cancer. Nature, 555(7697), 446-447. https://doi.org/10.1038/d41586-018-02881-7
Hashimoto, D. A., Rosman, G., Rus, D., & Meireles, O. R. (2018). Artificial intelligence in surgery: Promises and perils. Annals of Surgery, 268(1), 70-76. https://doi.org/10.1097/SLA.0000000000002693

No competing interests reported.

Download PDF

Reviewers invited by journal
21 Jul, 2024
Editor assigned by journal
14 Jul, 2024
Submission checks completed at journal
12 Jul, 2024
First submitted to journal
11 Jul, 2024

You are reading this latest preprint version

Prediction of Early Mortality in Patients Undergoing Right Hemicolectomy Using Machine Learning Algorithms

Status:

Version 1

Abstract

Figures

INTRODUCTION

METHODS

RESULTS

Demographics and Clinical Characteristics

Co-Morbidities

Surgical Indications and Outcomes

DISCUSSION

CONCLUSIONS

Declarations

References

Additional Declarations

Status:

Version 1