Heart Attack Prediction

Using Machine Learning

This presentation outlines the development of a machine learning model for
predicting heart attacks. The model analyzes a comprehensive set of clinical
parameters to determine the likelihood of a patient experiencing a heart attack.

Submitted By: Project Guide:

Isha Raghav Ms. Anandpreet Kaur
Kajal
Km.Khushbu
Introduction: Importance of Early Heart Attack Detection
1 Time is Critical 2 Improved Quality of Life
Early detection and intervention are crucial for improving patient Early diagnosis helps patients manage risk factors like hypertension
outcomes and minimizing the risk of complications. or cholesterol, leading to healthier lifestyles and better overall well-
being.

3 Cost-Effective Care
Preventing a heart attack through early detection reduces the need for expensive emergency treatments and long-term hospital stays.
Mission Objectives

Accuracy User-Friendly Interface Early Detection Scalability

Achieve a high level of accuracy Develop an intuitive and easy- Enable early detection of heart Ensure the model can be scaled
in predicting heart attack risk, to-use interface for healthcare attack risks, allowing for timely to handle larger datasets and
minimizing false positives and providers, enabling them to intervention and improving can be adapted for use in
false negatives to ensure input patient information patient outcomes. various healthcare settings,
reliable and trustworthy effortlessly and obtain accurate both clinical and remote.
predictions. predictions quickly.
Dataset Overview: Multivariate Clinical Parameters
Multivariate Clinical Parameters Data Insights and Predictive Relevance

The dataset includes 14 key health-related factors such as age, By studying these factors, we can find patterns related to heart
gender, chest pain type, blood pressure, cholesterol levels, blood attacks. For example, high cholesterol, high blood pressure, and
sugar, heart rate, and exercise-related changes. These factors help certain chest pain types are strongly linked to higher risk. These
provide a complete picture of a person’s heart health and are used patterns help our model make accurate predictions and identify
to predict the risk of a heart attack. people at risk early.
Algorithms Used: Random Forest, Decision Tree,
XGBClassifier
Random Forest Decision Tree XGBClassifier
• An ensemble method that • A simple tree-like structure that • A gradient boosting algorithm that
combines multiple decision trees to splits data into specific categories builds decision trees sequentially,
improve accuracy and reduce using a series of decision rules. with each tree correcting the errors
overfitting. • Easy to understand and interpret of the previous one.
• Works by averaging the predictions but prone to overfitting on complex • Known for its high performance
from individual trees to make datasets. and ability to handle complex
robust and reliable predictions. datasets efficiently.
Model Training and Evaluation Methodology

Data Collection 1
A comprehensive dataset containing 13 key clinical parameters was
assembled. These features are essential for predicting the likelihood of a
heart attack. 2 Data Preprocessing
The dataset was thoroughly cleaned by handling missing values, outliers,
and ensuring data consistency. Feature selection techniques were applied to
Model Training 3 identify the most relevant parameters for accurate prediction.
Three machine learning models were trained: Decision Tree, Random Forest,
and XGBClassifier. These models were trained using the preprocessed data
and evaluated for performance. 4 Model Evaluation
The models' performance was evaluated using metrics like AUC (Area Under
Curve) score and accuracy. Hyperparameters were fine-tuned to optimize the
Prediction and Outcome 5 models for better performance.
The best performing model was selected for real-time predictions. This
allows for early detection and intervention based on clinical parameters.
6 Results Interpretation
Insights from the models, such as feature importance and correlations,
helped in understanding key risk factors for heart attacks. This knowledge
guides preventive healthcare strategies.
Model Performance:
86.88% Accuracy

86.88%
Accuracy
The model correctly predicted heart attacks in 86.88% of cases.
Future Enhancements
The future work of this project will focus on developing a wearable device that
integrates sensors to collect real-time health data, such as heart rate, blood
pressure, and oxygen levels. This device will continuously monitor the user’s
vital signs and send the data to a machine learning model. The model will then
process this real-time information to predict the likelihood of a heart attack,
offering instant and accurate results.

Through this approach, users will have greater control over their heart health,
making informed decisions and receiving early warnings about potential heart
risks.
Benefits of Future Enhancement
Integrating real-time health data monitoring via a wearable device offers several key advantages:

Early and Continuous Portability and Convenience Low Power Consumption

Detection and Efficient Latency
The lightweight, portable device allows
Real-time monitoring enables early users to monitor heart health anytime, Designed for low power consumption
detection of potential heart risks, anywhere, enabling continuous and minimal latency, ensuring long-term
improving timely intervention and monitoring outside clinical settings. use and quick data processing/alerts.
preventing attacks before they occur.

Cost-Effective and Accessible Simplified and User-Friendly

Affordable devices will make heart health monitoring accessible Easy-to-use design eliminates complexity, making it accessible to
to a wider population, improving healthcare reach, particularly in all ages and backgrounds.
underserved areas.
Conclusion and Key Takeaways
1 Enhanced Accuracy: 2 Faster Decision-Making:

The developed machine By predicting heart attack risk

learning model improves the early, the model allows
precision of heart attack healthcare professionals to act
predictions, offering better risk quickly, potentially preventing
assessment and reducing life-threatening events and
diagnostic errors. saving lives.

3 Impact on Public Health:

By improving early detection and intervention, this project has the

potential to reduce the incidence and impact of heart attacks,
contributing to better public health outcomes.
Thank you