assa: Sniffing out diseases with technology

Overview

In the realm of medical diagnostics, a new approach is emerging—one that harnesses the power of scent to detect diseases before visible symptoms appear. Our project aims to develop an Advanced Smell Sensor Array (ASSA) capable of identifying distinct Volatile Organic Compound (VOC) profiles associated with various diseases, including Parkinson's and certain cancers.

Inspired by the remarkable ability of a woman in the UK to smell Parkinson's disease, we're working to automate and enhance this process using VOC sensing technology. By combining low-cost, high-accuracy sensors with machine learning techniques, we're striving to create a non-invasive, accessible tool for early disease detection that could transform preventative healthcare and significantly improve patient outcomes.

Objectives

Our project's primary goal is to develop a smell sensor array that can differentiate between various VOC patterns in the air, effectively distinguishing between different scents and potentially identifying disease-specific profiles. Specifically, we aim to:

1. Design and construct a low-cost, high-accuracy sensor array capable of detecting and differentiating various VOCs.
2. Implement machine learning techniques for pattern recognition and smell identification.
3. Validate the sensor array's effectiveness in detecting specific VOC profiles, with a particular focus on those associated with Parkinson's disease and certain cancers.
4. Create a functional prototype that can be further developed for clinical applications in early disease detection.

The primary beneficiaries of this project will be patients at risk of developing these diseases. By enabling earlier detection, we hope to improve treatment outcomes and quality of life. Additionally, healthcare providers and researchers in the field of disease diagnosis could benefit from this new diagnostic tool.

Methodology

Currently, our project is in the initial research phase. We've completed an extensive literature review focusing on disease-specific VOC profiles, existing electronic nose technologies, and various sensors suitable for VOC detection. Based on this research, we've identified promising sensor types and potential array configurations.

Our next steps will involve:

1. Sensor Selection and Array Design: We'll carefully select a range of Metal Oxide Sensors (MOS) and other relevant sensors based on their sensitivity to specific VOCs associated with our target diseases.
2. Prototype Construction: Using the components obtained from RS, we'll construct our first prototype. This will involve:
   • Setting up the sensor array on breadboards
   • Integrating environmental sensors to account for temperature and humidity
   • Connecting the sensors to microcontrollers (Arduino/Raspberry Pi) for data collection
   • Implementing analog-to-digital converters for accurate signal processing
   • Designing and integrating an air intake and suction mechanism
3. Data Collection and Processing: We'll develop software to collect and process the sensor data, converting raw signals into meaningful VOC profile data.
4. Machine Learning Model Development: Using collected data, we'll train machine learning models to recognize and classify different scent profiles.
5. Testing and Validation: Initially, we'll test the system's ability to differentiate between two individuals based on scent alone. If successful, we'll move on to more complex scenarios, potentially including tests with different perfumes to simulate more challenging conditions.
6. Iterative Improvement: Based on test results, we'll refine our sensor array, data processing algorithms, and machine learning models.

Product

With the funding prize from RS, we plan to purchase the following key components:

1. Metal Oxide Sensors: A variety of MOS sensors sensitive to different VOCs will form the core of our sensor array.
2. Environmental Sensors: Temperature and humidity sensors to account for environmental factors affecting VOC readings.
3. Microcontrollers: Arduino or Raspberry Pi boards for data collection and initial processing.
4. Breadboards and Electronic Components: Including jumper wires, resistors, capacitors, and op-amps for building sensor circuits and signal conditioning.
5. Analog-to-Digital Converters: To ensure accurate digitization of sensor outputs.
6. Voltage Regulators and Power Management ICs: For stable and efficient power supply to the system.
7. Enclosure: A basic ABS enclosure to house and protect our prototype during testing.
8. Air Pump and Tubing: For controlled sample delivery to the sensors, if budget allows.

These components will be crucial in constructing our initial prototype and conducting preliminary tests to validate our concept.

Next Steps

Our immediate next steps include:

1. Receiving and organizing the components from RS (expected by end of September).
2. Constructing the initial prototype (aiming for completion by November).
3. Developing and implementing the data collection and processing software.
4. Conducting initial tests to differentiate between two individuals based on scent.

We anticipate several challenges:

1. Sensor Sensitivity and Selectivity: Ensuring our sensors can detect the specific VOCs of interest at the required concentrations.
2. Environmental Interference: Dealing with confounding smells from the environment during training and operation.
3. Data Processing and Pattern Recognition: Developing robust algorithms to accurately identify specific VOC profiles amidst background noise.

We would consider our project successful if we can reliably differentiate between two individuals based on their scent profiles. This would serve as a proof of concept for the technology's potential in disease detection. Long-term success would involve detecting specific disease profiles and validating the results in a clinical setting.

Future steps beyond this initial phase would include collaborating with medical professionals, conducting more extensive testing, and potentially seeking regulatory approval for clinical use.

See the full list of Winners of this year's Student Project Fund - South Africa!