Home Humanities Language Computer Science Math Physics STEM I STEM II

STEM II

STEM II is a class taught by Dr. Crowthers. In STEM II, we build an assistive technology device.

Mid-Infrared Spectroscopy Device for Malaria Diagnosis with Machine Learning

Problem Statement

Malaria is prevalent and on the rise, particularly in more rural areas, where access to early detection and treatment may not be available. Particularly, from 2019 to 2020, malaria deaths increased by 10%. Evidently, Malaria has become a pressing issue for global health. Effective and prompt diagnosis is crucial to controlling malaria in rural and urban areas. As stated by the National Institute of Health, late diagnosis is the primary cause of death due to malaria (Tangpukdee et al., 2009).

Often, access to diagnostic methods is lacking in rural areas (Tangpukdee et al., 2009). Moreover, those in rural areas may not be able to make extensive commutes to their nearest healthcare provider. This is due to the mental and physical hardships that come with leaving the area. When suffering from symptoms, the challenge to leave their home becomes even larger. Additionally, people suffering from disabilities face mental and physical hardships as well.

Design Approach

However, spectroscopy provides a promising solution for detection. Spectroscopy is a chemical field of study where molecules excited via light, emit a spectra which is then recorded. Each element and molecule has its own unique spectra, allowing scientists to detect functional groups from the spectrometer graphs. In a spectrometer, a light is passed through a small hole or slit in a metal plate to isolate it. It is then bounced off of a grating to split the light, which is read by a detector (NASA, 2022). Particularly, mid-infrared spectroscopy has been seen in area of malaria detection. Near-infrared spectroscopy pushes light at different wavelengths and measures how much light was absorbed by the different bonds. It uses this information to plot the percent transmittance against the wavelength to then identify functional groups (Bennett et al.).

Typical devices in the field are static in the sense that the parts are non-mobile, and this constricts the devices to a laboratory setting with required high-end technology. This makes it hard for a general population to gain easy access to this kind of technology when needed. By innovating this technology into a portable device, it will be easier for people in target areas to reach out for early detection, in a cost-friendly, and portable way.

The objective is to create a device that will release different wavelengths of light that will excite the molecules. The spectra created by these excited molecules will then be seen through the device. This spectra was malaria specific and included key identifying markers for the disease. Addtionally, we implemented a machine learning model to differentiate malaia positive and negative patients. Logistic regression was used to create a binary classification model for diagnosis of malaria vs. non-malaria patients.

Initial Prototype

Final Design