Ethical AI Governance

Slide 1: Title Slide

Very Good morning everyone! I’m Achal Patil.

And I’m Mayuri Dudhe.

Together, we will present on a very important topic — Ethical AI Governance: Building

Responsible Technology for the Future.

Slide 2: Content
Here’s what we will cover today —
We will talk about what Ethical AI Governance is, why AI ethics matter, key principles, real-
world challenges, and more.

Slide 3: Introduction of Ethical AI Governance

Ethical AI Governance means building AI systems that are safe, transparent, and respect
human values.
_____________________________________________________________________________________________

Slide 4
To do this, we have some key components:

 Risk testing to avoid bias.

 Strong regulations, like AI laws.
 Ethical review boards to check AI usage.

And at each stage — design, testing, and deployment — we need to actively work to keep AI
ethics.
Slide 5: Definition and Importance of AI Ethics

So what is AI ethics? It means making sure AI is fair, responsible, and respects human rights.

 It prevents problems like bias, unfair decisions, or privacy violations.

 In short, it helps AI benefit everyone equally.

Slide 6: Key Ethical Principles in AI

Some key principles are:

 Fairness — treat everyone equally.

 Transparency — AI decisions should be explainable.
 Accountability — someone must be responsible for what AI does.
 Inclusivity — include diverse perspectives in AI design.

Slide 7: Role of Education and Advocacy

Education is key. We need to include AI ethics in schools and colleges so future developers
think responsibly. Also, advocacy—raising voices for ethical policies—is important. It helps
shape better laws and systems.

Slide 8: Real-World Challenges

- But there are real problems too. AI can be biased, like facial recognition misidentifying
people from minority groups.

- And AI-generated fake news can spread rapidly, creating chaos.

- Sometimes, it’s hard to know who to blame when AI makes a mistake—that’s the
accountability issue.

Slide 9: Advantages & Disadvantages

- Ethical AI builds public trust and avoids harm. But on the downside, it can slow down
innovation and increase costs.

- Yes, ethical checks take time—but they help prevent bigger problems later.

Slide 10: Conclusion

- To sum up, ethical AI isn’t just an option—it’s a necessity.

- By being transparent, fair, and responsible, we can make sure AI works for people, not
against them.

- Governments, companies, and all of us must come together to guide AI in the right
direction.

1. What is Ethical AI Governance?

A system of rules to ensure AI is fair, transparent, and responsible.

2. Why is AI Ethics important?

To prevent bias, protect rights, and build trust in AI.

3. Key principles of ethical AI?

Fairness, transparency, accountability, privacy, and inclusivity.

4. How to implement ethical AI?

Include diverse views, test for bias, monitor after deployment.

5. Real-world challenges?

Bias, fake info, unclear responsibility, and lack of transparency.

6. Role of education?

Teaches ethics, builds awareness, supports responsible AI use.

7. Advantages of ethical AI?

Trust, fairness, safety, and social acceptance.

8. Disadvantages?

Slower progress, hard to explain, complex regulations.

9. How to make AI transparent?

Use explainable AI and ethical review boards.

10. Who is responsible?

Everyone—govt, industry, and society must work together.

11. What is bias in AI?

Unfair treatment or decisions caused by biased data or design.

12. What is explainable AI?

AI that clearly shows how it made a decision.

13. What is the EU AI Act?

A law that sets strict rules for using AI in Europe.

14. Why is accountability important in AI?

So someone is responsible if AI causes harm.

15. How can AI affect jobs?

It can automate tasks and reduce human jobs.

16. Can AI make ethical decisions?

Not completely—humans must guide AI ethics.

17. What is the role of advocacy in AI ethics?

To push for fair policies and raise awareness.

18. Why is data privacy important in AI?

It protects people’s personal and sensitive data.

19. What happens if AI is not governed ethically?

It may cause harm, unfair results, and loss of trust.

20. How can students contribute to ethical AI?

By learning ethics, reporting issues, and building fair systems.

-----------------------------------------------------------------------------------------------------------------
-----------------------------------------------------------------------------------------------------------------
 Heart Health Monitoring System

Slide 1: Title Slide

Good morning everyone. Today we are here to present our project titled "Heart Health
Monitoring System". This project was developed under the guidance of Prof. A.
Nachankar at K.D.K. College of Engineering, Nagpur.

Slide 2: Content

Here is an overview of our presentation. We'll start with the introduction, followed by
heart attack basics, symptoms, objectives, system design, and finally outputs,
advantages, and future scope.

Slide 3: Introduction

Heart attacks are one of the leading causes of death globally. So early detection is very
important. Our system uses machine learning to predict the risk of a heart attack early.
We’ve used Python, Tkinter for GUI, and MySQL for database support to make the tool
easy to use and efficient.

Slide 4: What is Heart Attack?

A heart attack occurs when the blood flow to the heart gets blocked. This is mostly due
to the build-up of cholesterol and fat called plaques. When these plaques burst, they
can form clots which stop the blood flow and damage the heart muscle.

Slide 5: Symptoms

 Some common symptoms include chest pain, tiredness, nausea, shortness of breath,
and sweating.

Many times, people ignore these early signs. Our system aims to catch the risk
before it becomes dangerous.
Slide 6: Objective

Our main goal is to create a system that helps people know their heart health status by
analyzing simple inputs. We also wanted to build a tool that’s easy for users—patients
or doctors—to understand and use.

Slide 7-8: Requirements

 We used Python 3.12, Visual Studio, and MySQL. The system can run on Windows,
Linux, or Mac.

 We kept the hardware simple—8GB RAM, i5 processor—so that most laptops or

PCs can support it easily.

Slide 9: Machine Learning Algorithm

We used Logistic Regression, a basic but effective machine learning algorithm for
binary classification: heart disease or no heart disease. It takes user data—like age,
cholesterol level, and more—and gives a result based on that.

Slide 10: System Design

Here you can see the system design. It includes data input, model prediction, report
generation, and data storage.

The modular design makes the system easy to manage and update in the future.

Slide 11: Output Screens

These are some screenshots from our system – including the Sign Up page, Login page,
and Main Analysis page. Reports can also be printed or saved for future use—useful in
hospitals and clinics.
Slide 12: Advantages

Our system predicts heart attack risk early and uses a simple interface with Tkinter. It’s
fast, modular, stores data efficiently, and is suitable for educational and academic use.

Disadvantages:

However, logistic regression has its limits—it may not detect very complex patterns like
deep learning can.
Also, manual data entry can cause input mistakes, and we still need to improve data
security.

Slide 14: Future Scope

We can improve the system by adding wearable device integration—like

smartwatches that track heart rate in real time. We also see this being used in
telemedicine, genetic analysis, and long-term health tracking.

Slide 15: Conclusion

This project shows that with machine learning and basic tools, we can build a system
that really helps. It’s a step toward smart and digital healthcare.There are still
challenges like data privacy and model accuracy, but with continuous improvement, this
can become a very helpful tool.

Slide 16: References

These are the references we used while developing the system—from research papers
to technical guides.

Thank you all for listening! We’re happy to answer any questions now.
QUESTIONS CAN BE ASK……………..

General Questions

1. What is the main objective of your project?

The main objective is to develop a system that predicts the risk of a heart attack using
machine learning techniques based on patient health parameters.

2. Why did you choose heart attack prediction as your topic?

Heart disease is a leading cause of death globally. Early prediction can save lives. This
project aims to assist doctors and patients in timely diagnosis and prevention.

3. What datasets did you use and why?

We used a publicly available heart disease dataset (such as from Kaggle or UCI
repository) because it includes multiple relevant attributes like age, blood pressure,
cholesterol, and more.

4. Can you briefly explain the overall workflow of your system?

The workflow includes data collection, preprocessing, training a machine learning
model, creating a GUI using Tkinter, taking user input, predicting risk, and showing
visual results.

Technical Questions

5. Which machine learning algorithm did you use and why?

We used Logistic Regression because it performs well with binary classification tasks
like predicting presence or absence of heart disease.

6. How did you handle missing or noisy data?

We checked for missing values and either removed or imputed them using mean/median
values. Noisy data was filtered during preprocessing.
7. What features (parameters) are most influential in predicting heart attacks?
Parameters like age, sex, chest pain type, resting blood pressure, cholesterol, fasting
blood sugar, and maximum heart rate are key predictors.

8. How did you evaluate your model's performance? Which metrics did you use?
We used accuracy, precision, recall, and F1 score to evaluate the model. These metrics
help balance false positives and false negatives.

9. What was your model’s accuracy and how does it compare to existing systems?
Our model achieved around 85-90% accuracy, which is comparable or better than
similar machine learning models on the same dataset.

Implementation Questions

10. Which programming language and tools did you use?

We used Python with libraries like pandas, scikit-learn, matplotlib, and Tkinter for
GUI.

11. How is the GUI designed? Which library was used?

The GUI is created using Tkinter. It allows users to input health parameters, visualize
results, and view prediction reports.

12. How do you store patient data? Is there any backend database used?
We used MySQL for storing patient details and their results securely.

13. Is your system scalable to large datasets or real-time data from wearables?
Currently, it's a prototype, but it can be scaled to integrate real-time data from wearable
devices with additional processing capabilities.

14. Can the user print or download the prediction report from the system?
Yes, the system includes a print feature that prints the prediction result along with
graphs and patient details.
Ethical and Security Questions

15. How do you ensure patient data privacy in your system?

Patient data is stored securely in a local database. In future versions, we plan to add data
encryption and authentication features.

16. What are the potential ethical concerns with AI in healthcare that your system
addresses?
Our system avoids bias by using balanced datasets. It supports decision-making, not
replacing medical professionals, thus maintaining ethical use of AI.

Visualization and Reporting

17. What kind of visualizations are generated in your system and what insights do they
provide?
The system generates bar charts, pie charts, and line graphs showing patient
parameter trends and prediction distributions to help interpret results.

18. Can you explain one of your graphs and its relevance to prediction results?
For example, the pie chart shows the ratio of predicted positive vs negative heart
attack risk, giving a quick overview of the model’s output across users.

Future Scope and Innovation

19. How can your system be improved in the future?

- Add more advanced models like Random Forest or Deep Learning

- Enable cloud storage and access

- Integrate real-time data from fitness bands or smartwatches

20. Have you considered integrating wearable health data for real-time predictions?
Yes, it's part of our future scope. Real-time heart rate, BP, and ECG data from
wearables can enhance prediction accuracy and alert users instantly.

Can this model be used in real hospitals?

→ With validation, data security, and regulatory approval, it has potential
for real-world clinical use.

Can you integrate this system with mobile devices?

→ Yes, by converting the backend logic into a mobile app using
frameworks like Kivy or Flutter with Python APIs.

How do you ensure data privacy in your system?

→ By using MySQL for controlled data access and implementing proper
data handling practices.

What are the main risk factors for heart attack considered?
→ Age, gender, cholesterol, blood pressure, diabetes, obesity, smoking, and family history.

Why is early prediction of heart attack important?

→ It can save lives by enabling early medical intervention and lifestyle changes.

What are common symptoms of heart attack included in the system?

→ Chest pain, fatigue, shortness of breath, dizziness, and radiating pain.

What is the accuracy of your model?

→ Accuracy is calculated using accuracy_score, and it’s evaluated on both
training and test sets (specific value depends on dataset).

· What data did you use for training the model?

→ The publicly available heart.csv dataset, containing patient health indicators.

· · How did you handle missing values or data preprocessing?

→ Checked and managed missing values using Pandas, performed feature selection and
scaling as needed.

· · How did you split the dataset?

→ Used train_test_split() from scikit-learn to divide the data into training and testing
sets.

· · What is the accuracy of your model?

→ Accuracy is calculated using accuracy_score, and it’s evaluated on both training and test
sets (specific value depends on dataset).

· · What libraries did you use for the machine learning model?
→ sklearn for ML, numpy, pandas for data handling, and matplotlib for visualization.