National University of Technology

Computer Science Department (A)

Semester Fall – 2024
Assignment: 01
Course: Artificial Intelligence

SUBMITTED BY SUBMITTED TO
Name: Mahaz Abbasi Mam Noushin Saba
Reg no.: F23605035 CS Department (A)
 Assignment NO:01
Solution:
1.Briefly explain the case study and the motivation of using AI in healthcare.

AI in Healthcare:
This case study explores the implementation of an Intelligent Medical Diagnostic System (IMDS)
using Artificial Intelligence (AI) for the early detection and diagnosis of breast cancer. The system is
designed to analyze mammograms and other relevant medical data to improve diagnostic accuracy
and efficiency.
During a routine mammogram, Emily’s radiologist notes a suspicious mass in her left breast. The
radiologist refers her to the IMDS for further analysis.

Data Collection:
The IMDS collects and processes the following data:
Mammogram images: High-resolution images taken during Emily's recent screening.
Clinical history: Family history of breast cancer, personal medical history, and any previous imaging
results.
Physical examination results: Findings from Emily’s clinical examination, including breast
tenderness or changes.

AI Analysis:
Image Processing: The IMDS utilizes advanced deep learning algorithms to analyze the
mammogram images. It compares the images against a vast database of annotated mammograms,
looking for patterns indicative of malignancy.
Feature Extraction: The system identifies key features such as masses, calcifications, and
architectural distortions. It uses machine learning techniques to assess the size, shape, and density of
these features.
Risk Assessment: Based on its analysis, the IMDS calculates the likelihood of the mass being
malignant. It considers additional factors, such as Emily's family history and age, to refine its risk
assessment.
Results:
The IMDS produces a report indicating:
Diagnosis Probability: The AI system predicts a 75% likelihood that the detected mass is malignant
based on historical data and its analysis.
Recommended Next Steps: The system suggests further diagnostic procedures, including a breast
ultrasound and a biopsy to confirm the diagnosis.
Motivation for Using AI in Healthcare:
The motivation for integrating AI in healthcare, particularly in diagnostic systems,
includes:

1. Improved Diagnostic Accuracy: AI can analyze vast amounts of data more

accurately than humans, reducing diagnostic errors and improving patient outcomes.

2. Speed and Efficiency: AI systems can process medical data quickly, enabling faster
diagnosis and treatment decisions, which is crucial in emergencies.

3. Handling Complex Data: AI systems can process and integrate various types of
medical data, including imaging, lab results, and patient history, leading to more
comprehensive insights.

4. Supporting Healthcare Providers: By reducing the cognitive load on doctors, AI

systems can help manage large patient volumes, particularly in settings with limited
medical personnel.

5. Continuous Learning: AI-based diagnostic systems can improve over time by

learning from new data and research, keeping recommendations updated and aligned
with the latest medical standards.

AI in healthcare aims to enhance the quality, efficiency, and accessibility of medical

services, ultimately leading to better patient care.

2.IMDS Health Data Collection and Classification:

The Intelligent Medical Diagnostic System (IMDS) is a healthcare technology platform that uses AI,
machine learning, and data analytics to diagnose, monitor, and manage patient health by collecting
and processing vast medical data, providing accurate insights, disease progression predictions, and
personalized treatment recommendations.
Percepts:
• Vital Signs: Heart rate, blood pressure, temperature, respiratory rate, oxygen saturation.
• Medical Imaging: X-rays, CT scans, MRIs, ultrasound images.
• Laboratory Results: Blood tests, urine tests, cholesterol levels, glucose levels, etc.
• Electronic Health Records (EHRs): Medical history, medication records, past treatments,
allergies, and surgeries.
• Patient-Generated Data: Wearable health devices like fitness trackers.
• Symptoms Reports: Data from patient-reported symptoms.
• Genetic Information: DNA sequencing data indicating predispositions to certain conditions.
• Environmental Data: Data on air quality, pollutants, etc.

Health States:
• Healthy State: Normal vital signs, lab results, no abnormalities.
• At-Risk State: Identifies early warning signs of conditions.
• Acute Condition State: Detects immediate health concerns.
• Chronic Condition State: Recognizes long-term health issues.
• Critical/Deteriorating State: Flags patients showing rapid decline or severe health crises.
• Recovery State: Monitors for relapse or complications during recovery.
• End-of-Life State: Identifies and monitors the end-of-life stage in cases of terminal illness.

3.Design the PEAS framework for IMDS :

In Artificial Intelligence (AI), various types of agents operate to achieve specific goals. The PEAS
system is a critical framework used to categorize these agents based on their performance,
environment, actuators, and sensors. Understanding the PEAS system is essential for grasping how
different AI agents function effectively in diverse environments. Among these agents, Rational
Agents are considered the most efficient, consistently choosing the optimal path for maximum
efficiency.
AI System Design and Environment Understanding

Performance Measure
• Defines success of an agent based on precepts.
• A quantitative measure evaluating outcomes of actions against a predefined goal.
• Guides the agent's decision-making process.
• In self-driving cars, performance measure includes safety, efficiency, and comfort criteria.

Environment
• The surrounding environment of an agent.
• Five major types of environments: Fully Observable & Partially Observable, Episodic & Sequential,
Static & Dynamic, Distinct & Continuous, Deterministic & Stochastic.
• Environment includes external factors and conditions that the agent must consider.
Actuator
• Delivers the output of action to the environment.
• Responsible for executing actions based on perceptions and decisions.
• Design and choice of actuators directly affect the agent's ability to perform its functions.

Sensors
• Receptive parts of an agent that collect data from the environment.
• Provide necessary information for the agent to act intelligently.
• In autonomous drones, sensors include cameras, GPS, and altimeters.
• Quality and variety of sensors greatly influence the agent's ability to perceive and understand its
environment.

PEAS Framework for Medical Diagnostic System:

Performance Measure:
• Diagnostic Accuracy: System's ability to diagnose conditions and minimize false
positives/negatives.
• Response Time: System's speed in processing patient data and providing diagnostic outputs.
• Patient Outcomes: System's effectiveness in improving patient health through accurate diagnosis
and timely intervention.
• Prediction Accuracy: System's ability to predict disease progression or future complications.
• Efficiency: Reduction in time and effort required for healthcare professionals to diagnose
conditions.
• User Satisfaction: Positive feedback from healthcare providers using the system.
• Cost Reduction: Lowering healthcare costs by reducing unnecessary tests, optimizing resource
allocation, and improving overall efficiency.
• Regulatory Compliance: Adherence to medical regulations and standards.

Environment :
• Hospitals and Clinics: Assists doctors in patient diagnosis, tracking, and decision-making based on
medical data.
• Remote Healthcare: Used in telemedicine or home healthcare settings.
• Laboratories: Interacts with lab data, analyzes test results, and feeds them into the diagnostic
process.
• Wearable Devices: Operates in conjunction with wearable health tech providing continuous health
data.
• Multimodal Data Environments: Handles data from multiple sources such as medical imaging, lab
reports, patient history, and real-time monitoring systems.

Actuators :
• Diagnosis Recommendations: Suggests possible diagnoses based on patient data.
• Alerts and Notifications: Generates real-time alerts when abnormal vitals, potential health risks, or
life-threatening conditions are detected.
• Treatment Recommendations: Suggests personalized treatment plans based on the diagnosis and
patient history.
• Predictive Alerts: Sends out warnings about potential complications or disease progression based
on the patient's current health trajectory.
• Data Visualization: Provides visualizations of patient health data.
• Monitoring and Adjusting: Suggests adjustments in patient monitoring based on real-time data.
• Documentation: Automates the documentation of diagnostic processes and results for electronic
health records (EHRs).
Sensor:

• Vital Signs Monitors: Input: Data on heart rate, blood pressure, body temperature, oxygen
saturation, and respiratory rate.
• Medical Imaging Devices: Input: X-rays, MRI scans, CT scans, and ultrasound images.
• Laboratory Test Results: Input: Blood tests, urine tests, biopsy results, and genetic testing.
• Electronic Health Records : Input: Patient history, medication records, allergies, and prior
diagnoses.
• Wearable Health Devices: Input: Real-time data from patient-worn devices.
• Symptom Reporting: Input: Patient-reported symptoms.
• Genetic Data: Input: DNA sequencing or genetic tests identifying hereditary risks for certain
diseases.

4.Artificial Intelligence Environment Types:

Fully Observable vs Partially Observable

• Fully observable environment: The agent's sensor can sense or access the agent's complete state at
each point in time.
• Partially observable environment: The agent has no sensors in all environments.

Deterministic vs Stochastic
• Deterministic environment: The agent's current state completely determines the next state.
• Stochastic environment: Random environment, not unique and cannot be completely determined by
the agent.

Competitive vs Collaborative
• Competitive environment: The agent competes against another agent to optimize the output.
• Collaborative environment: Multiple agents cooperate to produce the desired output.

Single-agent vs Multi-agent
• Single-agent environment: The environment consists of only one agent.
• Multi-agent environment: The environment involving more than one agent.

Dynamic vs Static
• Dynamic environment: The environment constantly changes when the agent is up with some action.
• Static environment: The environment with no change in its state.

Discrete vs Continuous
• Discrete environment: The environment consists of a finite number of actions that can be
deliberated to obtain the output.
• Continuous environment: The environment in which the actions are performed cannot be numbered.

Episodic vs Sequential
• Episodic task environment: Each of the agent’s actions is divided into atomic incidents or episodes.
• Sequential environment: The previous decisions can affect all future decisions.

Known vs Unknown
• In a known environment, the output for all probable actions is given.
• In an unknown environment, an agent needs to gain knowledge about how the environment works.
Diagnostic System Classification:
Static or Dynamic:
- Operates in a dynamic environment due to evolving patient conditions.
- Requires continuous updates of diagnosis or recommendations.

Deterministic or Stochastic:
- Involves uncertainty in medical diagnoses.
- Outcomes can be influenced by unknown factors.

Episodic or Sequential:
- Operates in a sequential environment.
- Every diagnosis and treatment recommendation can affect a patient's future health.

Fully Observable or Partially Observable:

- Operates with incomplete information, limiting informed decision-making.

Discrete or Continuous:
- Deals with continuous data, requiring real-time monitoring and decision-making.

• Single-Agent or Multi-Agent:
- Functions in a multi-agent environment, interacting with other systems and healthcare providers.

5.Intelligent Medical Diagnosis System (IMDS) Classification:

Medical Diagnostic System Agent (IMDS) Functions

Model-Based Reflex Agent:

• The IMDS tracks a patient's medical history, symptoms, and treatments over time.
• It incorporates patient tracking, disease models, and treatment feedback to make informed
diagnoses.

Goal-Based Agent:
• The IMDS aims to improve patient outcomes, reduce diagnosis time, and prevent complications.
• It prioritizes diagnostic tasks and treatment plans based on current health state and medical
history.
Utility-Based Agent :
• The IMDS evaluates various diagnostic and treatment options, focusing on risk-benefit analysis
and resource optimization.
• It ensures decisions lead to the highest overall value for the patient's health.

Learning Agent :
• If the IMDS includes machine learning capabilities, it can improve over time by learning from its
previous experiences.
• It refines its diagnostic algorithms, adjusts its decision-making process based on feedback from
outcomes, and adapts to new data.

Not a Simple Reflex Agent:

• The IMDS considers past patient data, disease progression models, and treatment effectiveness,
allowing it to handle complex medical conditions.

6.Medical Diagnostic System (IMDS) Analysis:

Thinking Humanly:
• The IMDS mimics human reasoning by following diagnostic protocols similar to physicians.
• It lacks human-like cognitive abilities such as intuition, emotional insight, and subjective judgment.
• It operates purely through programmed rules and algorithms, mimicking logical thought processes
without the complexity of real human thinking.

Acting Humanly:
• The IMDS can act humanly to some extent by performing tasks commonly done by human doctors.
• It does not simulate human interaction in the same way a doctor might.
• It provides objective, data-driven recommendations, without the interpersonal warmth or bedside
manner of a human doctor.

Thinking Rationally:
• The IMDS excels at thinking rationally by using logical reasoning, knowledge-based processes, and
data-driven algorithms to arrive at the most accurate diagnosis or optimal treatment.
• It processes patient data and uses it to draw conclusions about the patient's condition.
• It evaluates all possible outcomes logically, applying medical knowledge and guidelines to assess
the likelihood of different diagnoses and treatments.

N 9/
• The IMDS takes actions designed to maximize the patient's health outcomes based on the data it
has.
• It suggests evidence-based treatment options that have the highest probability of success given the
patient's current condition, medical history, and test results.
• It ensures rational decision-making by using utility-based approaches to select treatments or
diagnoses that maximize expected patient benefits.