Module 12
Artificial Intelligence (AI) in Healthcare
Start
Special Acknowledgement
Section 1: Overview of Artificial Intelligence
Section 2: Artificial Intelligence in Medicine
Index
Section 3: Ethical Oversight of AI
Overview of Artificial Intelligence (AI)
Objectives
Section 1: Overview of Artificial Intelligence
By the end of this section, you will be able to:
Differentiate Artificial Intelligence (AI) from Generative AI
List and define at least 5 key-terms related to artificial intelligence
OBJECTIVES
Define and describe AI Hallucinations
Describe at least 5 best practices for Prompting
Artificial Intelligence Definitions:
Artificial Intelligence
Definition 1
Definition 2
Brief overview
Definition 3
Definition 4
What is Artificial Intelligence?
Subfields: - machine learning - deep learning - rule-based systems
Use this side of the card to provide more information about a topic. Focus on one concept. Make learning and communication more efficient.
Use this side of the card to provide more information about a topic. Focus on one concept. Make learning and communication more efficient.
Use this side of the card to provide more information about a topic. Focus on one concept. Make learning and communication more efficient.
Use this side of the card to provide more information about a topic. Focus on one concept. Make learning and communication more efficient.
Use this side of the card to provide more information about a topic. Focus on one concept. Make learning and communication more efficient.
Use this side of the card to provide more information about a topic. Focus on one concept. Make learning and communication more efficient.
Initiated from computer science & mathematics
Goal is to mock human intelligence
AI vs. Generative AI
Various definitions
Title
Flip through the different aspects.
Write a brief description here
Brief Historical Context to AI
1997
2011
1956
IBM's Deep Blue in World Chess ChampionshipExample of how machine reasoning and computization can outperform a human cognitive task
Dartmouth Conference Defined as the founding of AI and John McCarthy developed the term "AI"
IBM Watson wins "Jeopardy!"Computational power to answer questions (National Language Processing)
Brief Historical Context to AI
2022
2016
OpenAI launches ChatGPT AI available for public and professional usage, an increase in questions about ethics and hallucinations
Google DeepMind developed an AI algortithm AlphaGo defeats Lee Sedol Performance in non-linear tasks, inspired nonpattern recognition
Overview of AI
Data: The "Fuel" of AI
What is it?
Special Considerations
Why it Matters
Purpose
- Data is the input (“fuel”) that AI systems use to learn and find patterns.
- It can be structured (e.g., lab values) or unstructured (e.g., clinical notes, images).
- To enable AI to analyze large datasets and support predictions or decisions.
- AI performance depends on data quality (Garbage In, Garbage Out).
- Big data challenges (variety, velocity, veracity, value) affect AI reliability.
- Big data challenges (variety, velocity, veracity, value) affect AI performance.
- Incomplete, biased, or inaccurate data can lead to unreliable or unfair outcomes.
- AI insights can influence clinical decisions and patient outcomes.
- Data quality directly impacts safety, equity, and trust in AI systems.
Example in Medicine
Hover over button
Goal: Transparency and Trust
AI Levels
Artificial Intelligence(AI)
Interpretability
Machine Learning
eXplainability (XAI)
Deep Learning
AI: Training the Model
2. Clean, Prepare, & Manipulate Data
1. Get Data
4. Test Data
3. Train Model
5. Improve
AI OVERVIEW
Types of Model Training
Neural Networks
Supervised Learning
- Use layers of interconnected nodes inspired by the human brain.
- Power many modern AI systems, including imaging and language models.
Unsupervised Learning
- Learns from labeled data with known outcomes.
- Identifies patterns to predict or classify new data (e.g., normal vs. pneumonia X-rays).
- Works with unlabeled data.
- Finds hidden patterns, relationships, or clusters (e.g., grouping patients with similar symptoms).
Natural Language Processing (NLP)
Reinforcement Learning
- Enables AI to understand and analyze human language.
- Commonly used for clinical notes, reports, and other text-based medical data.
- Learns through trial and error using feedback.
- Improves decisions over time based on rewards (e.g., treatment planning simulations).
Machine Learning (ML) - Deep Leanring (DL)
(provides archictecture for pattern recognition)
GenAI Systems: Producing Content
Natural Language Processing (NLP)
(applies DL to language understanding)
Large Language Model (LLM)
(trained on large text datasets)
Generative AI
What is the process?
(creates new text, image, audio)
Prompt
(human instruction that triggers generation)
Output new knowledge or communication
Review:AI Explained: AI, Machine Learning, Deep Learning, and Gen AI" (6 min)
The "c's" of Human Intelligence & AI
Human Intelligence Movement
Collaboration
Communication
Creativity
Critical Thinking
Curiosity
AI OVERVIEW
Oversight of AI
Biases, Causes & Risks
AI Hallucinations: Causes & Risks
Hallucination
Bias
- When AI generates incorrect or fabricated information that appears accurate.
- Can spread misinformation, especially if outputs are not fact-checked.
- Occurs when AI is trained on unrepresentative or skewed data.
- Can lead to inaccurate or unfair outcomes that do not reflect reality.
Responsible AI / Governance
Grounding (RAG)
- Involves oversight structures (e.g., governance boards) to monitor AI use.
- Addresses ethical, legal, and accountability concerns in AI systems.
- Retrieval-Augmented Generation (RAG) reduces hallucinations by grounding responses in sources.
- The model retrieves relevant information before generating an answer.
AI OVERVIEW
AI prompting & Verification
BEST PRACTICES
Common Techniques
- Use specific language, defined roles, and clear framing.
- Break complex tasks into step-by-step instructions.
- Include constraints such as format, length, or citations.
Purpose
What is it?
To guide the model toward reliable, task-appropriate responses. To reduce ambiguity, errors, and irrelevant outputs.
- AI prompting is the practice of giving clear, structured instructions to a Large Language Model (LLM).
- More detailed prompts improve the accuracy and relevance of AI outputs.
Special Considerations
Ground prompts in trusted sources when possible to improve accuracy. Validate and iterate using human oversight (human-in-the-loop) before applying results.
Example in Medicine
Hover over button
Let's Review: AI LITERACY
Algorithm
Data
AI
Tuning
Info
Info
Info
Info
Machine Learning (ML)
Deep Learning (DL)
Neural Networks
Inference
Info
Info
Info
Info
Rule Based Systems
Model
Training
Supervised Learning
Info
Info
Info
Info
Let's Review: AI LITERACY
Hallucinations
Bias
Unsupervised Learning
Grounding (RAG)
Info
Info
Info
Info
Reinforcement Learning
Prompting
Human-in-the-Loop
Transfer Learning
Info
Info
Info
Info
Large Language Model (LLM)
Generative AI
Responsible AI
AI Agents
Info
Info
Info
Info
Knowledge Check
AI In medicine
Objectives
Section 2: AI in Medicine
By the end of this section, you will be able to:
Explain why Artificial Intelligence is important to healthcare
Provide examples of how AI is being used to support healthcare
OBJECTIVES
List three factors that influence AI adoption in healthcare systems
Describe how AI is supporting maternal health
Review:How AI is Revolutionizing Medicine (7:50 min)
AI and Healthcare Timeline
Brief Overview
Click on the + button to read more!
1980s
DXplain
1970s
INTERNIST-1
2000s
ML and EHRs
Radiological Image and Pathology & computers
1970s
2015
Watson System
1972
MYCIN
2023
Responsible AI Frameworks
AI IN MEDICINE
Why does AI matter to Healthcare?
Common Techniques
Special Considerations
What is it?
Purpose
- Used to simplify high-dimensional data into fewer, interpretable dimensions
Used to examine multidimensional relationships
Example in Medicine
Hover over button
Why does AI Matter to Healthcare?
AI in Remote Monitoring and Telehealth
AI in Clinical Decision Support
AI in Social Science Research
Click on the + to read more.
AI in Epidemiology
Factors Influencing AI Adoption in Medicine
Future of healthcare big data
Provides Trust & Liability Concerns
Integration into Clinical Workflow
Data Sensitivity & Privacy
Clinical Validation & Safety
Read more
Read more
Read more
Read more
Ethical & Bias Concerns
Interdiscplinary Collaboration
Patient Acceptance & Digital Equity
Read more
Read more
Read more
AI In Maternal Health Case Studies
Case Study 1
Article: Diagnostic Accuracy of an Integrated AI Tool to Estimate Gestational Age From Blind Ultrasound Sweeps. JAMA. 2024 Aug 27;332(8):649-657.
Key Take-Aways:
- AI-based Ultrasounds to estimate Gestational Age (GA)
- Ability for AI to support low-resource settings
- Potential for AI expansion in maternal health and help strengthen equity
Case Study 2
Article: On AI approaches for promoting maternal and neonatal health. Frontiers in Public Health, 10, 880034.
AI Maternal Health Opportunities
Technology Challenges
- Databases & Representation
- Trustworthy Model & Privacy
- Workflow Integration
Knowledge Check
Oversight of AI
Objectives
Section 3: Oversight of AI
By the end of this section, you will be able to:
Define the types of AI Bias
Describe why AI ethics is important to modern medicine
OBJECTIVES
Describe the importance of AI Governance
Describe AI and future needs
OVERSIGHT OF AI
Bias
Click on the + to read more.
Ethics & Medical AI
AI Governance & Strategy
AI OVERSIGHTS
JAMA: AI in Medicine
AI Tools
Evaluation
Clinical, Consumer, Business Operations, Hybrid
- Which AI Tools need Evaluation?
- How should it be evaluated?
- Who should be responsible to evaluate?
“AI, Health, Health Care Today and Tomorrow” (Angus, 2025)
Ensure AI is Equitable to improve health outcomes
COMPLETED:
SUMMARY
- Describe the importance of statistical literary and medicine
- Differentiate between various statistical techniques
- Define various types of Statistical Methods
Section 1: Overview of Statistical Methods
- Describe why data cleaning is important to the research process
- Define some common data assumptions that statisticians check for
- List common methods used for missing data & outliier
- Describe common steps taken for Data Assurance & Reproducability
Section 2: Data Cleaning, Missing Data & Outliers
Section 3: Best Practices for Results & Discussion/Conclusion
- Defind what is a statistical p-value
- Differentiate between association vs. causation, confidence interval & effect sizes
- Describe the difference between clinical and statistical significance
- List best practices & common erros regarding interpratation of results, discussion & conclusion sections
REFERENCES
REFERENCES
MODULE 12 COMPLETED
Remember to review what you've learned!
Threats
Contextualize your topic
- Plan the structure of your communication.
- Give it a hierarchy and give visual weight to the main point.
- Add secondary messages with interactivity.
- Establish a flow through the content.
- Measure results.
Opportunities
Contextualize your topic
- Plan the structure of your communication.
- Give it a hierarchy and give visual weight to the main point.
- Add secondary messages with interactivity.
- Establish a flow through the content.
- Measure results.
Smart Data
Ultimately, it is not about just having a lot of data or big data, it is about how to use this data for impactful, actionable, equitable decisions that add ‘value’ to patients and communities. It is working to move from Big Data (to) Smart Data.
AI & Big Data
AI is now assisting with how medical doctors diagnose, treat patients, and provide preventative care. However, each user of AI technology should be aware of AI’s limitations and provide curated human insights to suggestions provided.
Interoperability
The ability to integrate data from many sources calls for the need to be interoperable. Interoperable means that data can be shared and interpreted using shared data standards. Two terms you might hear about are: FHIR (pronounced as FIRE, it stands for Fast Healthcare Interoperability Resources) and Open APIS (APIs is Application Programming Interfaces) which allows for roadmaps and plans to exchange data based on standards.
Federated Learning
Many institutions, to balance privacy and innovation, are moving toward Federated Learning whereby AI models are trained on Decentralized data and no need to ‘move’ the data across institutions. The big takeaway is that the technology is ever rapidly evolving and it is important to stay abreast of emerging technologies. Always question - how will it benefit and what are potential unintended consequences too with the ultimate goal of Do no harm.
Strengths
Contextualize your topic
- Plan the structure of your communication.
- Give it a hierarchy and give visual weight to the main point.
- Add secondary messages with interactivity.
- Establish a flow through the content.
- Measure results.
Weaknesses
Contextualize your topic
- Plan the structure of your communication.
- Give it a hierarchy and give visual weight to the main point.
- Add secondary messages with interactivity.
- Establish a flow through the content.
- Measure results.
Federated Learning
Many institutions, to balance privacy and innovation, are moving toward Federated Learning whereby AI models are trained on Decentralized data and no need to ‘move’ the data across institutions. The big takeaway is that the technology is ever rapidly evolving and it is important to stay abreast of emerging technologies. Always question - how will it benefit and what are potential unintended consequences too with the ultimate goal of Do no harm.
Ethics, Privacy, and Patient Consent
Before using or contributing to Big Data, have you stopped to think about how data is being reused? Are patients being informed fully on how their individual data might be contributed to a larger sum of data? Just as one obtains consent for treatment same should be true and transparent for the contributions of individual data towards aggregate data to assist with population health level insights.
Smart Data
Ultimately, it is not about just having a lot of data or big data, it is about how to use this data for impactful, actionable, equitable decisions that add ‘value’ to patients and communities. It is working to move from Big Data (to) Smart Data.
Module 12
Emily Sheehy
Created on January 6, 2026
Start designing with a free template
Discover more than 1500 professional designs like these:
View
Essential Business Proposal
View
Project Roadmap Timeline
View
Step-by-Step Timeline: How to Develop an Idea
View
Artificial Intelligence History Timeline
View
Mobile Phone Call
View
Momentum: Tools Tutorial
View
Momentum: Onboarding Video
Explore all templates
Transcript
Module 12
Artificial Intelligence (AI) in Healthcare
Start
Special Acknowledgement
Section 1: Overview of Artificial Intelligence
Section 2: Artificial Intelligence in Medicine
Index
Section 3: Ethical Oversight of AI
Overview of Artificial Intelligence (AI)
Objectives
Section 1: Overview of Artificial Intelligence
By the end of this section, you will be able to:
Differentiate Artificial Intelligence (AI) from Generative AI
List and define at least 5 key-terms related to artificial intelligence
OBJECTIVES
Define and describe AI Hallucinations
Describe at least 5 best practices for Prompting
Artificial Intelligence Definitions:
Artificial Intelligence
Definition 1
Definition 2
Brief overview
Definition 3
Definition 4
What is Artificial Intelligence?
Subfields: - machine learning - deep learning - rule-based systems
Use this side of the card to provide more information about a topic. Focus on one concept. Make learning and communication more efficient.
Use this side of the card to provide more information about a topic. Focus on one concept. Make learning and communication more efficient.
Use this side of the card to provide more information about a topic. Focus on one concept. Make learning and communication more efficient.
Use this side of the card to provide more information about a topic. Focus on one concept. Make learning and communication more efficient.
Use this side of the card to provide more information about a topic. Focus on one concept. Make learning and communication more efficient.
Use this side of the card to provide more information about a topic. Focus on one concept. Make learning and communication more efficient.
Initiated from computer science & mathematics
Goal is to mock human intelligence
AI vs. Generative AI
Various definitions
Title
Flip through the different aspects.
Write a brief description here
Brief Historical Context to AI
1997
2011
1956
IBM's Deep Blue in World Chess ChampionshipExample of how machine reasoning and computization can outperform a human cognitive task
Dartmouth Conference Defined as the founding of AI and John McCarthy developed the term "AI"
IBM Watson wins "Jeopardy!"Computational power to answer questions (National Language Processing)
Brief Historical Context to AI
2022
2016
OpenAI launches ChatGPT AI available for public and professional usage, an increase in questions about ethics and hallucinations
Google DeepMind developed an AI algortithm AlphaGo defeats Lee Sedol Performance in non-linear tasks, inspired nonpattern recognition
Overview of AI
Data: The "Fuel" of AI
What is it?
Special Considerations
Why it Matters
Purpose
Example in Medicine
Hover over button
Goal: Transparency and Trust
AI Levels
Artificial Intelligence(AI)
Interpretability
Machine Learning
eXplainability (XAI)
Deep Learning
AI: Training the Model
2. Clean, Prepare, & Manipulate Data
1. Get Data
4. Test Data
3. Train Model
5. Improve
AI OVERVIEW
Types of Model Training
Neural Networks
Supervised Learning
Unsupervised Learning
Natural Language Processing (NLP)
Reinforcement Learning
Machine Learning (ML) - Deep Leanring (DL)
(provides archictecture for pattern recognition)
GenAI Systems: Producing Content
Natural Language Processing (NLP)
(applies DL to language understanding)
Large Language Model (LLM)
(trained on large text datasets)
Generative AI
What is the process?
(creates new text, image, audio)
Prompt
(human instruction that triggers generation)
Output new knowledge or communication
Review:AI Explained: AI, Machine Learning, Deep Learning, and Gen AI" (6 min)
The "c's" of Human Intelligence & AI
Human Intelligence Movement
Collaboration
Communication
Creativity
Critical Thinking
Curiosity
AI OVERVIEW
Oversight of AI
Biases, Causes & Risks
AI Hallucinations: Causes & Risks
Hallucination
Bias
Responsible AI / Governance
Grounding (RAG)
AI OVERVIEW
AI prompting & Verification
BEST PRACTICES
Common Techniques
Purpose
What is it?
To guide the model toward reliable, task-appropriate responses. To reduce ambiguity, errors, and irrelevant outputs.
Special Considerations
Ground prompts in trusted sources when possible to improve accuracy. Validate and iterate using human oversight (human-in-the-loop) before applying results.
Example in Medicine
Hover over button
Let's Review: AI LITERACY
Algorithm
Data
AI
Tuning
Info
Info
Info
Info
Machine Learning (ML)
Deep Learning (DL)
Neural Networks
Inference
Info
Info
Info
Info
Rule Based Systems
Model
Training
Supervised Learning
Info
Info
Info
Info
Let's Review: AI LITERACY
Hallucinations
Bias
Unsupervised Learning
Grounding (RAG)
Info
Info
Info
Info
Reinforcement Learning
Prompting
Human-in-the-Loop
Transfer Learning
Info
Info
Info
Info
Large Language Model (LLM)
Generative AI
Responsible AI
AI Agents
Info
Info
Info
Info
Knowledge Check
AI In medicine
Objectives
Section 2: AI in Medicine
By the end of this section, you will be able to:
Explain why Artificial Intelligence is important to healthcare
Provide examples of how AI is being used to support healthcare
OBJECTIVES
List three factors that influence AI adoption in healthcare systems
Describe how AI is supporting maternal health
Review:How AI is Revolutionizing Medicine (7:50 min)
AI and Healthcare Timeline
Brief Overview
Click on the + button to read more!
1980s
DXplain
1970s
INTERNIST-1
2000s
ML and EHRs
Radiological Image and Pathology & computers
1970s
2015
Watson System
1972
MYCIN
2023
Responsible AI Frameworks
AI IN MEDICINE
Why does AI matter to Healthcare?
Common Techniques
Special Considerations
What is it?
Purpose
Used to examine multidimensional relationships
Example in Medicine
Hover over button
Why does AI Matter to Healthcare?
AI in Remote Monitoring and Telehealth
AI in Clinical Decision Support
AI in Social Science Research
Click on the + to read more.
AI in Epidemiology
Factors Influencing AI Adoption in Medicine
Future of healthcare big data
Provides Trust & Liability Concerns
Integration into Clinical Workflow
Data Sensitivity & Privacy
Clinical Validation & Safety
Read more
Read more
Read more
Read more
Ethical & Bias Concerns
Interdiscplinary Collaboration
Patient Acceptance & Digital Equity
Read more
Read more
Read more
AI In Maternal Health Case Studies
Case Study 1
Article: Diagnostic Accuracy of an Integrated AI Tool to Estimate Gestational Age From Blind Ultrasound Sweeps. JAMA. 2024 Aug 27;332(8):649-657.
Key Take-Aways:
Case Study 2
Article: On AI approaches for promoting maternal and neonatal health. Frontiers in Public Health, 10, 880034.
AI Maternal Health Opportunities
- Predictive Analytics
Technology ChallengesKnowledge Check
Oversight of AI
Objectives
Section 3: Oversight of AI
By the end of this section, you will be able to:
Define the types of AI Bias
Describe why AI ethics is important to modern medicine
OBJECTIVES
Describe the importance of AI Governance
Describe AI and future needs
OVERSIGHT OF AI
Bias
Click on the + to read more.
Ethics & Medical AI
AI Governance & Strategy
AI OVERSIGHTS
JAMA: AI in Medicine
AI Tools
Evaluation
Clinical, Consumer, Business Operations, Hybrid
“AI, Health, Health Care Today and Tomorrow” (Angus, 2025)
Ensure AI is Equitable to improve health outcomes
COMPLETED:
SUMMARY
Section 1: Overview of Statistical Methods
Section 2: Data Cleaning, Missing Data & Outliers
Section 3: Best Practices for Results & Discussion/Conclusion
REFERENCES
REFERENCES
MODULE 12 COMPLETED
Remember to review what you've learned!
Threats
Contextualize your topic
Opportunities
Contextualize your topic
Smart Data
Ultimately, it is not about just having a lot of data or big data, it is about how to use this data for impactful, actionable, equitable decisions that add ‘value’ to patients and communities. It is working to move from Big Data (to) Smart Data.
AI & Big Data
AI is now assisting with how medical doctors diagnose, treat patients, and provide preventative care. However, each user of AI technology should be aware of AI’s limitations and provide curated human insights to suggestions provided.
Interoperability
The ability to integrate data from many sources calls for the need to be interoperable. Interoperable means that data can be shared and interpreted using shared data standards. Two terms you might hear about are: FHIR (pronounced as FIRE, it stands for Fast Healthcare Interoperability Resources) and Open APIS (APIs is Application Programming Interfaces) which allows for roadmaps and plans to exchange data based on standards.
Federated Learning
Many institutions, to balance privacy and innovation, are moving toward Federated Learning whereby AI models are trained on Decentralized data and no need to ‘move’ the data across institutions. The big takeaway is that the technology is ever rapidly evolving and it is important to stay abreast of emerging technologies. Always question - how will it benefit and what are potential unintended consequences too with the ultimate goal of Do no harm.
Strengths
Contextualize your topic
Weaknesses
Contextualize your topic
Federated Learning
Many institutions, to balance privacy and innovation, are moving toward Federated Learning whereby AI models are trained on Decentralized data and no need to ‘move’ the data across institutions. The big takeaway is that the technology is ever rapidly evolving and it is important to stay abreast of emerging technologies. Always question - how will it benefit and what are potential unintended consequences too with the ultimate goal of Do no harm.
Ethics, Privacy, and Patient Consent
Before using or contributing to Big Data, have you stopped to think about how data is being reused? Are patients being informed fully on how their individual data might be contributed to a larger sum of data? Just as one obtains consent for treatment same should be true and transparent for the contributions of individual data towards aggregate data to assist with population health level insights.
Smart Data
Ultimately, it is not about just having a lot of data or big data, it is about how to use this data for impactful, actionable, equitable decisions that add ‘value’ to patients and communities. It is working to move from Big Data (to) Smart Data.