1 of 14

​

​

​

​

​

184: Efficient classification and Recognition of the Fruit

Images using Hybrid Features and Grey Wolf

Optimization Algorithm

​

Harmandeep Singh1, Sumeet Kaur2

1Department of Computer Science, Mata Gujri Khalsa College, Kartarpur(Punjab)-144801

2Department of Economics, Amity University, Mohali

​

Track 1: Image Processing,Computer Vision and Pattern Recognition (ICP)

2 of 14

Outline of presentation

  1. Introduction
  2. Motivation and Background
  3. Literature Review
  4. Proposed Work
  5. Results
  6. Conclusion
  7. Future Scope
  8. References

184: Pushing the Boundaries of Mortality Prediction: Advancing High-Risk Sepsis-III Patient Care through Cutting-Edge Deep Learning Techniques

3 of 14

Introduction

Sepsis-III is a condition marked by physiological, pathological, and biochemical abnormalities triggered by infection.

    • Global Impact: In 2017, 49 million sepsis cases, causing 11 million fatalities, constituting nearly 20% of global deaths.
    • Vulnerable Demographics: Disproportionate impact on newborns, expectant mothers, and those in resource-limited environments.
    • WHO Definition: Critical medical condition with an excessive immune response, causing significant damage by various infectious agents.

​

​

​

​

184: Pushing the Boundaries of Mortality Prediction: Advancing High-Risk Sepsis-III Patient Care through Cutting-Edge Deep Learning Techniques

4 of 14

Introduction

184: Pushing the Boundaries of Mortality Prediction: Advancing High-Risk Sepsis-III Patient Care through Cutting-Edge Deep Learning Techniques

5 of 14

Introduction

​

Research Focus:

​

  • Clinical prediction model development for refined healthcare systems and improved preventive measures against sepsis.

​

  • Utilizing machine learning algorithms (XGBoost, LGBM, Logistic Regression, Linear SVC, and ANN) for mortality prediction.

​

​

​

​

184: Pushing the Boundaries of Mortality Prediction: Advancing High-Risk Sepsis-III Patient Care through Cutting-Edge Deep Learning Techniques

6 of 14

Motivation and Background

  • Addressing the challenges in mortality prediction for high-risk Sepsis-III patients involves navigating through the intricacies of a complex medical scenario.
  • The inherent difficulty lies in the dynamic and multifaceted nature of sepsis, particularly in its severe manifestations.
  • High-risk Sepsis-III patients represent a subset with elevated mortality risks, demanding a nuanced approach to prognosis.
  • Accurately predicting mortality in this context becomes a formidable task due to the interplay of various factors, including the rapid progression of organ dysfunction, diverse patient responses to infection, and the critical timeframe within which interventions must occur.
  • The unpredictable nature of sepsis progression further amplifies the challenge, as the severity of organ dysfunction may escalate swiftly, necessitating timely and precise interventions to improve patient outcomes.

184: Pushing the Boundaries of Mortality Prediction: Advancing High-Risk Sepsis-III Patient Care through Cutting-Edge Deep Learning Techniques

7 of 14

Literature Review

184: Pushing the Boundaries of Mortality Prediction: Advancing High-Risk Sepsis-III Patient Care through Cutting-Edge Deep Learning Techniques

8 of 14

Contributions

184: Pushing the Boundaries of Mortality Prediction: Advancing High-Risk Sepsis-III Patient Care through Cutting-Edge Deep Learning Techniques

The significant contributions of work are:

(i) Enhanced Mortality Prediction: This study contributes to the field by addressing the challenging problem of accurately predicting mortality in high-risk sepsis-III patients.

(ii) Comprehensive Comparative Analysis: This research provides a thorough comparative analysis of a wide range of machine learning and deep learning techniques.

(iii) Utilization of Key Metrics: This work employs essential evaluation metrics, including accuracy and AUC Score, to systematically identify the most effective model for mortality prediction.

9 of 14

Proposed Work

184: Pushing the Boundaries of Mortality Prediction: Advancing High-Risk Sepsis-III Patient Care through Cutting-Edge Deep Learning Techniques

10 of 14

Results

​

​

184: Pushing the Boundaries of Mortality Prediction: Advancing High-Risk Sepsis-III Patient Care through Cutting-Edge Deep Learning Techniques

11 of 14

Conclusion

  1. Successful achievement of predicting sepsis-3 patient mortality through various machine learning models.
  2. Comparative analysis reveals LGBM Classifier as the top-performing model, excelling in accuracy, AUC score, and F1 score compared to others.
  3. ANN model, while accurate, suggests potential for further optimization through hyperparameter tuning.
  4. Our research introduces a pioneering machine learning model with a significant impact on healthcare, particularly for sepsis-3 patients.
  5. Emphasis on predicting mortality, utilizing a range of critical health factors, positions our model as a valuable tool in improving patient outcomes.

184: Pushing the Boundaries of Mortality Prediction: Advancing High-Risk Sepsis-III Patient Care through Cutting-Edge Deep Learning Techniques

12 of 14

Future Scope

  1. Improvement : Emphasis on refining the ANN model and integrating PCA for improved performance.
  2. - Current ANN model accuracy stands at 0.8, prompting exploration of hyperparameter tuning for enhanced predictive capabilities.
  3. ​
  4. - Integration of Principal Component Analysis (PCA) as a dimensionality reduction technique to boost efficiency and reduce computational complexity.
  5. ​
  6. - Merging hyperparameter tuning and PCA to establish a comprehensive strategy for optimizing model performance.

​

​

  • Potential Application: Promises more effective healthcare strategies through early prediction, precise diagnosis, and comprehensive sepsis severity assessment.

​

  • Anticipated Impact: Redefining sepsis management, contributing to improved patient outcomes, and marking a significant advancement in healthcare.

​

184: Pushing the Boundaries of Mortality Prediction: Advancing High-Risk Sepsis-III Patient Care through Cutting-Edge Deep Learning Techniques

13 of 14

References

  1. Jaesung Yoo. (2022). Refined MIMIC-III 30-day mortality prediction of sepsis-3 patients. IEEE Dataport. https://dx.doi.org/10.21227/qrhd-a469.
  2. Hou, N., Li, M., He, L., Xie, B., Wang, L., Zhang, R., Yu, Y., Sun, X., Pan, Z., & Wang, K. (2020). Predicting 30-days mortality for MIMIC-III patients with sepsis-3: a machine learning approach using XGboost. Journal of translational medicine, 18(1), 462. https://doi.org/10.1186/s12967-020-02620-5.
  3. Singh, I., Singh, S. K., Singh, R., & Kumar, S. (2022, May). Efficient loop unrolling factor prediction algorithm using machine learning models. In 2022 3rd International Conference for Emerging Technology (INCET) (pp. 1-8). IEEE.
  4. Kline, A., Wang, H., Li, Y. et al. Multimodal machine learning in precision health: A scoping review. npj Digit. Med. 5, 171 (2022). https://doi.org/10.1038/s41746-022-00712-8
  5. Hasib, K. M., Towhid, N. A., & Islam, M. R. (2021). Hsdlm: a hybrid sampling with deep learning method for imbalanced data classification. International Journal of Cloud Applications and Computing (IJCAC), 11(4), 1-13.
  6. Wang, H., Li, Y., Naidech, A. et al. Comparison between machine learning methods for mortality prediction for sepsis patients with different social determinants. BMC Med Inform Decis Mak 22 (Suppl 2), 156 (2022). https://doi.org/10.1186/s12911-022-01871-0.
  7. Perng J-W, Kao I-H, Kung C-T, Hung S-C, Lai Y-H, Su C-M. Mortality Prediction of Septic Patients in the Emergency Department Based on Machine Learning. Journal of Clinical Medicine. 2019; 8(11):1906. https://doi.org/10.3390/jcm8111906.

​

​

​

184: Pushing the Boundaries of Mortality Prediction: Advancing High-Risk Sepsis-III Patient Care through Cutting-Edge Deep Learning Techniques

14 of 14

  • Queries ???

184: Pushing the Boundaries of Mortality Prediction: Advancing High-Risk Sepsis-III Patient Care through Cutting-Edge Deep Learning Techniques