AI in Research & Social Care

Moving Beyond the Chatbot

Mustafa Aslan

2025-04-24

Outline

  • What is AI?
  • AI in Research
  • Key Generative AI Tools for Researchers
  • Social Care Use Cases
  • Predicting Future Care Demand
  • Critical Considerations

What is AI?
(It’s not just ChatGPT/Gemini/Copilot)

AI is a Toolbox, Not a Magic Wand

Main Engines of AI

Most people see Generative AI, but the engine of research is often:

  • Machine Learning (ML): Teaching computers to learn from data to make predictions.
  • Forecasting: Predicting future trends based on historical data and patterns.
  • Optimization: Solving complex “best-fit” problems.
  • Natural Language Processing (NLP): Understanding and generating human language.
  • Computer Vision: Analyzing images and videos for insights.
  • Reinforcement Learning: Training agents to make decisions through trial and error.
  • Generative AI: Creating new content (text, images, code) based on learned patterns using large language models (LLMs).

AI in Research

AI as a Research Partner/Assistant, not a Replacement



“AI doesn’t replace the researcher; it removes the friction of the mundane.”

Key Applications

  1. Data Synthesis: Mapping 20 years of research in seconds.
  2. Pattern Discovery: Finding signals and patterns in complex datasets that humans might miss.
  3. Automated Workflows: Managing large-scale longitudinal studies.
  4. Hypothesis Generation: Suggesting new research directions based on existing data.

Key Generative AI Tools for Researchers


🔬 Perplexity AI

  • Real-time research synthesis with citations, fact-checking, and current information retrieval for literature review

📚 SciSpace (Typeset)

  • AI-powered paper analysis and understanding: break down complex papers, extract key findings, answer questions

📖 NotebookLM

  • Transform research documents into interactive conversations, study guides, and podcast-style discussions

🤖 ChatGPT/Claude/Gemini/Copilot

  • General-purpose AI assistants for brainstorming, explanation of concepts, coding and writing support

Social Care Use Cases

Social Care Use Cases



Social Care Use Cases

AI can transform the quality of life for both carers and residents:

AI tool Use Case Inputs (Data) Outputs (Result) Benefit
Machine Learning Fall Prevention Sensor data, gait history Real-time “High Risk” alert Early intervention
Forecasting Planning Demand Demographics, health trends 5-year demand projections Strategic planning
Optimization Staff Scheduling Staff skills, care plans Optimized shift rosters Matches skills to needs
NLP Voice Assistants Verbal commands Reminders, device control Independence
Vision AI Gait Analysis Video/Wearable motion Mobility trend reports Tracking rehab progress
Generative AI Documentation Raw notes, voice logs Structured care reports Efficiency boost

The Universal Logic of AI: \(X \rightarrow f(X) \rightarrow \hat{Y}\)

Example: Predicting Future Care Demand


\(X\)
📥 Inputs
  • Historical bed occupancy
  • New referrals per week
  • Demographics & diagnosis
  • Seasonality (holidays/flu)





\(\rightarrow\)

\(f(X)\)
📊 AI Model
  • The “Pattern Finder”
  • Learns the relationship between your inputs (\(X\)) to understand the “why” and “when.”





\(\rightarrow\)

\(\hat{Y}\)
📤 Outputs
  • Predicted occupancy
  • Confidence intervals
  • Risk alerts for staffing
  • Actionable planning insights


Researcher’s Translation: We take what we know (\(X\)), apply a logic/function (\(f\)), to get our best estimate of the future (\(\hat{Y}\)).

Example (continued): What are the “Ingredients” and the “Results”?


Context \(X\) (The Input) \(f(X)\) (The Logic) \(\hat{Y}\) (The Prediction)
Regression A resident’s age or medical history A mathematical trend line A Risk Score: e.g., 75% probability of a fall
Generative AI (ChatGPT/Gemini) The words in your prompt A massive Neural Network The Next Word: Summarizing
Forecasting (social care planning) Historical bed usage + Staffing levels Pattern-matching algorithm Future Demand: Number of beds needed in December


The “Researcher’s Secret”

Whether it’s a simple spreadsheet or a complex chatbot, the machine is always doing the same thing: taking what we know (X) to calculate a best guess (\(\hat{Y}\)).

Crucial for Qualitative Researchers: The AI calculates the probability, but it doesn’t understand the meaning. You provide the meaning.

Critical Considerations

Ethical and Practical Considerations for AI in Social Care



  • Transparency: Explainable AI for accountability
  • Privacy: Secure data handling and consent
  • Equity: Avoiding bias in vulnerable populations
  • Human-in-the-Loop: Final decisions remain with care professionals

AI is a Tool to Assist, Not Replace !!!