Artificial Intelligence and Data Analytics

AI can be categorized in several ways. Here are some of the most common types:

• 1. Narrow AI (Weak AI)
  • Designed for a specific task.
  • Examples: voice assistants and recommendation systems.
  • Most AI today falls into this category.
• 2. General AI (Strong AI)
  • A theoretical form of AI that can perform any intellectual task a human can.
  • Still under research and development.
• 3. Reactive Machines
  • These respond to specific inputs but don’t store memories (e.g., IBM’s Deep Blue chess computer).
• 4. Limited Memory
  • Uses past experiences to make future decisions (e.g., self-driving cars).
• 5. Theory of Mind and Self-Aware AI
  • Hypothetical future AI that understands emotions and has self-awareness.

AI systems learn using different approaches:

• 1. Supervised Learning
  • AI learns from labeled data (e.g., identifying diseases from X-rays).
    • An example is predicting diseases from X-rays and student performance.
• 2. Unsupervised Learning
  • Finds patterns in unlabeled data (e.g., grouping similar patients).
    • Example: Clustering students by learning styles.
• 3. Reinforcement Learning
  • Learning by trial and error with rewards or penalties.
    • Example: AI playing video games or optimizing classroom layouts.
• 4. Semi-Supervised and Self-Supervised Learning
  • AI combines small-labeled datasets with large unlabeled ones.
  • It’s useful when labeling data is expensive or time-consuming.

AI is the umbrella term, and several subfields fall under it. For example:

• 1. Machine Learning (ML)
  • Enables systems to learn from data.
    • Example: Predicting student dropout rates.
• 2. Deep Learning (DL)
  • A type of ML using neural networks with many layers.
    
    • Example: Image recognition in dental X-rays.
• 3. Natural Language Processing (NLP)
  • Enables machines to understand and generate human language.
    • Example: Chatbots, translation tools and sentiment analysis.
• 4. Generative AI (GenAI)
  • Creates new content (text, images, music).
    • Example: AI writing assistance, generating synthetic medical data.

AI is used for several general cases; however, it’s transforming both education and health care in powerful ways.

• General Applications:
  • Quickly condenses long articles, reports or meeting notes.
  • Revises text for grammar, tone and clarity.
  • Generates ideas and brainstorming topics, titles or creative content.
  • Supports presentations with outlines, talking points or slide content.
  • Translates language in real time.
  • Automates repetitive tasks like scheduling or data entry.
• In Education:
  • Personalizes learning tailored to student needs.
  • Automates grading and feedback.
  • Uses predictive analytics to identify at-risk students.
  • Offers chatbots and virtual tutors for 24/7 support.
• In Healthcare:
  • Uses medical imaging for detecting diseases in X-rays and MRIs.
  • Uses predictive diagnostics to forecast patient outcomes and risks.
  • Provides virtual health assistants to support patients.
  • Accelerates drug discovery and treatment development.

To ensure Artificial Intelligence (AI) is used effectively and responsibly across our institution, we recommend the following best practices:

• Be Transparent: Always disclose when used in academic, clinical or research settings.
• Verify Outputs: Review AI-generated content for accuracy, bias and relevance.
• Protect Data: Don’t share personal or sensitive data with public AI tools unless covered by a DUA (Data Use Agreement) or BAA (Business Associate Agreement).
• Use Ethically: Avoid using AI to deceive, plagiarize or misrepresent work.
• Be Aware of Bias: Remember, AI tools are as good as the data used for training it.
• Stay Informed: Keep up with institutional policies and updates.

At Texas Tech Health El Paso, ethics guides all AI applications. We are committed to:

• Fairness: Minimizing bias in data and models.
• Transparency: Ensuring AI systems are explainable.
• Accountability: Defining responsibility for AI-driven outcomes.
• Privacy: Protecting individual data rights.
• Compliance: Following institutional, state and federal ethics guidelines.

Evaluating AI models is essential to ensure they are fair and reliable. Here are key concepts used:

• Common Metrics
  • Accuracy: Percentage of correct predictions (best for balanced datasets).
  • Precision: How many predicted positives are correct.
  • Recall (Sensitivity): How many actual positives were identified.
  • F1 Score: Balance between precision and recall (useful for imbalanced data).
  • AUC-ROC: Measures the ability of a model to distinguish between classes.
• For Regression Models
  • Mean Absolute Error (MAE): The average of the absolute differences between predicted and actual values.
  • Mean Squared Error (MSE): The average of the squared differences between predicted and actual values.
  • R² Score (Coefficient of Determination): The proportion of variance in the dependent variable that is predictable from the independent variables.
• For Generative Models
  • BLEU / ROUGE: For evaluating text generation.
  • Fréchet Inception Distance (FID): For image generation quality.
  • Human Evaluation (with SMEs): Often necessary for subjective tasks like summarization or creativity.
• Fairness and Bias Checks
  • Evaluate performance across demographic groups.
  • Use tools like SHAP or LIME for explainability.
  • Align metrics with project goals (e.g., prioritize recall in medical diagnostics).