Research Prototype · TU/e Group 5

Non-invasive diabetes screening from saliva spectra

Upload an ATR-FTIR infrared spectrum file from a saliva sample and receive an instant machine-learning-based screening result for Type 2 diabetes risk.

This tool is a research proof-of-concept only. It does not provide medical diagnosis. Always consult a healthcare professional.
Wavenumber (cm⁻¹) — Sample --- Control 4000→400 cm⁻¹
1,040
Training samples
ATR-FTIR
Spectroscopy method
Binary
Classification task
Spectrum Analyzer

Upload your spectra file

Accepted formats: CSV (wavenumber, absorbance columns) or JSON array of absorbance values from 399–4000 cm⁻¹.

📂
Drop file here or click to browse
Your file never leaves your browser — no data is stored.
CSV JSON
📄
Screening Result
Low risk detected
Confidence score
Spectral points
Wavenumber range
Model
ANN · Optuna
Uploaded infrared spectrum
This result is generated by a research model trained on 1,040 ATR-FTIR saliva spectra. It is not a clinical diagnosis. Saliva composition is affected by diet, hydration, and oral health. Please consult a medical professional for proper testing.
The Process

How does it work?

From saliva sample to screening result in four steps.

1

Collect saliva

A small saliva sample is collected after fasting. No needles, no blood.

2

ATR-FTIR scan

The sample is scanned with infrared spectroscopy to capture its molecular fingerprint.

3

Upload spectrum

Export the spectrum as a CSV or JSON file and upload it here.

4

AI screening

Our neural network analyzes the spectral pattern and outputs a risk confidence score.

Type 2 Diabetes — Learn More

Understanding Type 2 Diabetes

🩸

What is Type 2 Diabetes?

Type 2 diabetes is a chronic condition where the body doesn't use insulin effectively, causing blood glucose levels to rise. It accounts for over 90% of all diabetes cases worldwide and affects more than 500 million people.

⚠️

Risk Factors

Key risk factors include being overweight, physical inactivity, family history of diabetes, age over 45, and a history of gestational diabetes. Lifestyle changes can significantly reduce risk.

🔬

Why Saliva?

Saliva contains biochemical information about metabolic changes in the body. ATR-FTIR spectroscopy can detect molecular-level differences between diabetic and non-diabetic saliva — offering a non-invasive alternative to blood tests.

🏃

Prevention

Regular physical activity (150 min/week), a balanced diet low in processed sugars, maintaining a healthy weight, and avoiding smoking can significantly lower your risk of developing Type 2 diabetes.

📊

About This Tool

This prototype was built as part of a TU/e Multi-Disciplinary CBL project. The neural network was trained on 1,040 ATR-FTIR saliva spectra from the SEDENA dataset (Mexico, 2019–2020). It is not a certified medical device.

🏥

Get Tested Properly

Official diabetes screening uses fasting blood glucose (≥126 mg/dL) or HbA1c (≥6.5%). If you're concerned, contact your GP. Early detection leads to much better outcomes.