ClinicTrends AI – Transforming Customer Feedback Into Intelligent Insights
As I transition from over 10 years in project management into life as a self-made software engineer, I’ve encountered a golden opportunity: designing and building a complete software project from scratch.
One of the core subjects I’m currently studying is Software Engineering Principles at Torrens University Australia, under the guidance of Dr. Ranju Mandal — a distinguished lecturer in Cybersecurity and a member of the Centre for Artificial Intelligence and Optimisation. With more than six years of postdoctoral research in AI and Big Data Analytics, Dr. Mandal brings invaluable expertise to both our coursework and the ClinicTrends AI project.
If that scenario doesn’t give you goosebumps… I’m not sure what will!
How ClinicTrends AI Was Born
During my previous career, I worked with a dataset of over one million customer records, including more than three years of NPS-style survey responses from multiple aesthetic clinics I helped manage. That dataset included over 25,000 individual responses. The more I studied software engineering and machine learning, the more I became obsessed with one question:
Can we turn raw customer feedback into actionable, real-time intelligence for businesses?
That’s how ClinicTrends AI was born — and it’s already progressing into version 2.0!
🚀 What’s Under the Hood?
We just wrapped up Release 1.0, which includes:
✅ A modular Streamlit web application architecture
✅ Implementation of four machine learning models for sentiment analysis:
- TF-IDF + Logistic Regression on comments
- Comment-score fusion models
- Hugging Face Transformers for context-aware predictions
✅ Multi-language support with automated translation
✅ Unit testing implemented via Pytest
🔮 What’s Coming Next
We’re actively developing Release 2.0, which will introduce:
- Real-time alerts for NPS score drops, enabling early intervention
- More explainable AI models for transparent sentiment predictions
- Enhanced feature engineering and hyperparameter optimization
- Future integrations like RESTful APIs and CRM connectivity
🎯 The Business Problem We’re Solving
Small and medium-sized businesses often rely on traditional NPS surveys that provide historical snapshots of customer satisfaction. But by the time negative trends are visible, customer churn may already be happening.
ClinicTrends AI changes the game by delivering:
- Real-time sentiment analysis across thousands of survey responses
- Predictive analytics to identify at-risk customers before they leave
- Multi-model comparisons so business leaders can choose the best-performing ML approach
- Cost savings compared to enterprise tools like Medallia or Qualtrics, which can cost over $50,000 annually
💻 Technical Stack
Here’s a quick peek at our tech stack:
| Component | Technology |
|---|---|
| Frontend | Streamlit |
| ML/NLP | Scikit-learn, Hugging Face Transformers, TextBlob |
| Data Processing | pandas, numpy |
| Visualizations | Altair, Plotly, matplotlib |
| Translation | deep-translator |
| Deployment | Streamlit Cloud (Docker-ready) |
🌐 Want to Explore?
If you’d like to check out the app:
- Live Demo: ClinicTrends AI Streamlit App
- GitHub Repo: ClinicTrends AI – GitHub
And if you’re curious about how Torrens University splits its trimester and assessments, I explained that in this article: My Journey at Torrens University.
🚀 Let’s Connect!
Building ClinicTrends AI has been the perfect bridge between my project management background and my software engineering future. I’d love to hear from fellow engineers, data scientists, or anyone passionate about transforming customer data into actionable insights.
Feel free to connect with me on LinkedIn or drop me a message!
Top comments (0)