Finsen Headphones: Treating Mid-Ear Infections Using Machine Learning and Phototherapy
Every year, there are 700 million cases of mid-ear infections (Otitis Media) and nearly 21,000 deaths worldwide. Many of those impacted are children and underprivileged populations. Without medical access and or healthcare, diagnosis and treatment of mid-ear infections are often difficult. Leanne’s invention, Finsen Headphones, provides a low-cost option to detect and treat a mid-ear infection using machine learning and blue light therapy – potentially preventing up to 60% of hearing loss in children.
A Novel MoS2 Electrode for Enhancing Electrochemical Hydrogen Production
Hydrogen (H2) can improve energy consumption by producing an alternative energy source to carbon-based fuels. It has numerous benefits, such as high energy density and zero pollution. However, the current method to produce hydrogen in industries uses fossil fuels, producing carbon dioxide (CO2). Electrochemical hydrogen production can be a clean and efficient alternative to producing hydrogen, with water being the only byproduct. There are several considerations for electrochemical hydrogen production like cathode material, electrolytes, and applied voltage. In this research, metal electrodes (e.g., Zn, Cu, Al, Ni, Sn, and Fe) were tested as metal catalysts for electrochemical hydrogen production to determine optimum operational conditions (e.g., types and concentrations of electrolytes).
OVision - The Automatic Assessment of Ovarian Cancer Features & Mesothelin Protein Overexpression from Histopathological Images using Deep Learning
Ovarian cancer (OC) is the deadliest cancer of the female reproductive organs. 65% of ovarian cancer patients die each year in the United States alone. The best way to reduce that number is for doctors to develop a treatment plan that is specific to the patient and their condition. However, to provide patients with the best diagnosis, doctors need to accurately classify the OC subtypes: High-Grade Serous Carcinoma, Low-Grade Serous Carcinoma, Endometrioid Carcinoma, Mucinous Carcinoma, and Clear Cell Carcinoma. By simply looking at histopathological images, it is often difficult for pathologists to distinguish between the subtypes accurately. Samaira’s research project utilizes a Deep Learning Convolutional Neural Network, based on VGG-16, to classify ovarian cancer subtypes and detect whether mesothelin protein was overexpressed, based on histopathological images. Current platforms for uploading histopathological images and getting results on cancer types require a great deal of technical expertise. Aiming to change that, with this platform, any doctor can get access, and receive immediate, accurate results.
Video Analysis of Hand Tremor to aid in Telehealth
Hand tremor disorders are frequently misdiagnosed, due to a lack of accurate and reliable measurements outside of hospital lab settings. In addition, access to neurologists can be a challenge for rural populations. To address these challenges, Sritej created a Python code utilizing OpenCV and MediaPipe libraries to analyze computer vision measurements of hand movement from video capture. Based on his study, Sritej concluded that video analysis of hand tremors provides accurate and reliable measures and would potentially reduce tremor misdiagnosis, help remotely and effectively manage patient care, and aid in telehealth.
ASD Screener - Early Risk Assessment of Autism Spectrum disorder using Machine Learning
Autism is one of the most common developmental disabilities that impact social and communication skills. While there is no cure, intensive early intervention can make a big difference in the quality of life for many children. Although it varies between individuals, the gut microbiome has helped successfully predict illnesses. Since microbiome data is high dimensional and non-linear, Amritha researched various feature reduction algorithms to develop an intelligent model that would be efficient and help in Autism Spectrum Disorder (ASD) prediction with more accuracy. This model can help physicians diagnose ASD at an early age using gut microbiome data from fecal samples. This model will also allow researchers to quickly and accurately detect microbial signatures that can serve as biomarkers of ASD risk.
Pura Aerem: Leveraging Catalytic Converters for Enhanced Filtering Efficacy
Air pollution is one of the greatest environmental risks to health. By reducing air pollution levels, countries can reduce the burden of diseases from stroke, heart disease, lung cancer, and both chronic and acute respiratory diseases, including asthma. Shanza Sami has developed a five-stage prototype with a diffuse filtration system, carbon nanotube (CNT) screening, electrolysis chamber, hydrogen fuel cells, and photoelectrochemical oxidation (PECO) technology. The carbon nanotube filtration system will be utilized to adhere to carbon dioxide through dynamic simulations. Additionally, water vapor will be utilized through an electrolysis chamber and hydrogen fuel cell to separate water vapor into breathable oxygen and hydrogen. Shanza hopes to use PECO technology to decompose volatile organic compounds that are dangerous to inhale.
Soil Fungi Arbuscular mycorrhizal fungi Fights Climate Change using Artificial Intelligence and IOT
Carbon dioxide (CO2) is the main greenhouse gas responsible for rising global temperatures and abrupt climate changes which threaten food production and ecological systems. Soils help manage climate change and contain two to three times more carbon than the atmosphere. A natural fertilizer, mycorrhizal fungi can increase carbon storage by improving the productivity of crops and vegetation. Using Artificial Intelligence and the Internet of Things (IoT), Saharsa created a method to analyze the soil used for crops that can help predict early crop growth and disease detection. Her solution utilizes three soil moisture sensors, a PH sensor, and a temperature sensor to monitor different soil conditions for plant growth. Using regenerative agricultural practices with eco-friendly AMF (soil-borne fungi) can promote a natural method of capturing carbon dioxide in a cost-effective way of removing CO2 from the atmosphere, which helps in carbon sequestration.
Piezoelectric Power Generation from Automotive Tires
Global warming, climate change, and air pollution are some of the most prevalent global issues of today. High levels of carbon dioxide and other greenhouse gasses in our atmosphere create concern for the health of our environment. The transportation sector is the largest contributor to U.S. greenhouse gas emissions, many of which come from passenger cars. To reduce greenhouse gas emissions from passenger cars, the use of electric vehicles needs to grow, but the drawback for many is the limited power of electric vehicle use. To help solve this issue, Asvini invented a way to generate clean power while driving to increase the distance an electric vehicle can travel before needing to be charged. Asvini’s invention uses piezo material on automotive tires to generate clean, environmentally friendly power that will recharge a car while driving. She hopes her invention will attract more people to buy electric vehicles and help reduce the carbon footprint associated with driving.
A Novel Dosing Pump to Prevent Clogs and Organic Overgrowth In AC Condensate Lines
Air conditioner condensate drain lines can clog every six to 12 months, costing millions in heating, ventilation, and air conditioning (HVAC) service calls and property damages. This issue can also cause severe health problems for the elderly and asthmatics due to organic overgrowths such as mold and Legionella bacteria. To address this issue, Daniel developed a solution using diluted chlorine to prevent overgrowth, a peristaltic pump, and an ESP8266 Wi-Fi programmer to time the dispensary of the chlorine. Daniel hopes his technique will produce fewer clogs and organic overgrowth in AC condensate lines, which will save homeowners and business money and reduce exposure to dangerous mold and bacteria.
The Comptometrist: An Efficient Way to Determine Myopic Power
The calculation of eye power can be slow and error-prone. On average, it takes nearly an hour to complete an eye examination which can open a window for error. Every year, more than 2.8 million people find significant issues with their optical prescription. To address this issue, Harini developed The Comptometrist, a prototype designed to cut down the time needed to determine myopic power in a patient’s eyes. Harini’s prototype would eliminate crowding in clinics, report accurate measures of myopic power in seconds, and closes the window of error in the eye examination process.