A Scalable and Freely Accessible Machine Learning Based Web Application for the Early Detection of Dyslexia
Isha Puri, Horace Greeley High School, NY
Dyslexia is the world’s most common neurological learning disability - it is estimated that it affects 1 in every 10 people worldwide. And though it has been proven that an early diagnosis can significantly reduce difficulties later in life, many families do not have access to affordable, accurate screenings for dyslexia. Currently, most screenings are either extremely expensive, costing around $1000-$2000, or require hefty scientific equipment, effectively rendering them inaccessible to a majority of the world. My goal is to build a free web-based application that will use a standard computer webcam to screen a child while reading and then use an accurate eye tracking algorithm to detect if the child has a higher risk of dyslexia. Using a novel combination of two different machine learning algorithms to detect the corners of the eye and and the center of the pupil, the eye tracker has a very high accuracy, with a maximum possible error of only a few pixels. These eye tracking results are then analyzed to determine the duration and frequency of gaze fixations made while reading, as studies have shown that dyslexics exhibit significantly longer and more frequent fixations. Based on this metric, the application will be able to predict if a child has a higher risk of dyslexia. Because it will be completely free and will not require any hefty scientific equipment, this solution will be accessible to millions of families around the world without regard to financial status.
Design and Prototype of a Virtual Music Teacher
Lauren Li, Westview High School, OR
Music enriches life, and can actually make us healthier. Playing music strengthens multiple areas of the brain and has even shown to delay the onset of dementia. However, learning an instrument can be extremely frustrating, not to mention the annoyance of turning music sheet pages. With extensive research already done on audio signal processing for music applications and new technologies such as augmented reality and speech/voice recognition, why not apply these technologies to music learning? My goal is to design an intelligent system that can recognize and track what and how a learner is playing against the music sheet, as well as provide useful feedback in real time. To demonstrate the feasibility, I will build a prototype that can automatically flip a page, and identify errors in notes, dynamics, and articulation. In this project, I will develop a simple language to describe sheet music in a format easily understood by computers. I will also develop and implement audio analysis algorithms to recognize four basic elements of music: melody, harmony, rhythm, and dynamics. After being able to track and compare the extracted musical elements against the computer representation of sheet music, I will create a page turner and feedback generator, with the outcome displayed on a monitor. In the end, I will play Minuet in G Major (BWM Anh. 114)—both perfectly and with errors—to test my prototype. I believe technology can make music education more accessible, so everyone can enjoy the benefits of learning a musical instrument.
Low-Cost Transradial Prosthesis Using Novel Methods for Optimizing Muscular Control
Zuriel Joven and Luke Torre, West Campus High School, CA
The amputation of one or both hands can significantly impair the capability of the amputee to perform daily activities. An amputation between the wrist and elbow isknown as a transradial amputation, as the amputation is made along the radial bone of the forearm. A transradial prosthetic arm can help the amputee regain some functionality by offering an emulation of the hand’s natural function. Unfortunately, prostheses can be unaffordable for many people, and those with lower price tags often come with less utility. Our proposal aims to eliminate this issue. We plan to build a low-cost transradial prosthesis equipped with dexterous control using a cost-effective method unused by state-of-the-art prostheses: binary force myography. The arm comprises 3D-printed parts, fishing line, Servo motors, an open-source circuit board, limit switches, and Velcro. After building the arm, we will dedicate our time to perfecting the control method through testing and iterating. The 8-grip five-finger prosthesis is controlled by limit switches fixed on the forearm muscles. Pattern recognition algorithms interpret the activation of different switches as different grip intentions. This novel control method could eliminate the price barrier of advanced transradial prostheses, as the product is projected to cost less than $50.
Smart Scheduling of Irrigation using Weather Sensor Data - IoT Applied for an Optimized Gravity Fed Drip Irrigation Method
Visala Tallavarjula, Wilcox High School, CA
Irrigation consumes more than 80% ofthe world’s fresh water. Exploding globalpopulation and increasing food demand leads to water shortage. Improvement of irrigation efficiency will help to mitigate water stress. My previous top soil bed method showed up to 25% evaporation loss reduction, and a low cost infiltration insert innovation enabled sub-surface like irrigation without needing expensive equipment. In addition, root zone texture/charcoal amendment engineering can lead to improved crop yield. Percolation Control layer can help to retain water especially for regions with sandy soils. These optimizations combined with gravity fed drip irrigation method along with smart irrigation scheduling can help small farm owners in the developing world, achieve water conservation. Charcoal amended rootzone layer preparation, placement of inserts at the emitters of the drip tube and perlite/peat-moss top soil bed can be set up at the farms. In addition, an IoT controlled smart valve can supply water from an elevated tank in precise amounts as needed by the plants. This minimizes water wastage and improves plant yield. An IoT system will consist of weather sensors to measure wind speed, relative humidity, ambient temperature, pressure and solar radiation. Every hour a real-time clock takes readings, a microcontroller unit (MCU) calculates the water loss due to evapotranspiration using weather sensors’ data, and a radio relays the information to a remote valve that can be opened for a predetermined duration using a separate MCU. Both MCUs and sensors will be solar powered for easy implementation in remote areas.
In Situ Repair of Bone Defects using Engineered Novel Photocrosslinkable Polymer Composite
Nipun Jayatissa, Maumee Valley Country Day School, OH
The regeneration of bone defects caused by trauma, fracture, and disease is a significant clinical challenge for patients in the United States and around the world. Annually, the cost for bone fracture repair exceeds $19 billion. Annual fractures and costs are projected to increase by 50% in 2025. Currently, there are several techniques including recent 3D printing to fabricate polymer or polymer composite scaffolds to use in scaffold guided bone tissue engineering. All of these scaffolds are pre-fabricated before implanting into the defect. These pre-fabricated scaffolds are not properly fit into the size of the bone defect. Up to date, there is no in situ bone repair material or technology. Therefore, engineering of in situ bone repair material and techniques will benefit the United States as well as the entire world. I plan to use simple but effective techniques to engineer in situ photocrosslinkable polymer composite (PCPC) gel for bone repair. This PCPC gel will be prepared using poly(ethylene glycol)dimethacrylate (PEGDMA), methyl cellulose, photoinitiator and nano-hyrdroxyapatite(nHA) as the main components. The prepared PCPC gel will be cured using a UV/visible light energy source. Next, hardened PCPC scaffolds will be characterized for morphology and pore architecture using Scanning Electron Microscope (SEM), compressive modulus with mechanical testing system, and cell viability and cell proliferation with murine osteoblasts. Finally, the validated PCPC gel will be injected to repair a rat femur defect.
Truck Tropical Storm: A Novel Method of Preventing Water-Related Road Accidents
Ryan Hanson, Janesville CSD, IA
Road crashes are ranked as the 9th leading cause of death globally. There are currently 1.2 billion estimated drivers worldwide, and there are hundreds of dangers a motorist can face on the roadway. Through extensive research, surveys, and expert testimonials, I have found water being sprayed off a semi’s tires during or after a rain to be a plausible problem as this water is sprayed onto another vehicles windshield creating poor visibility. There are existing products on the market that indirectly attempt to solve this problem and fail because it isn’t the main focus of that product. Windshield wipers are directly related to the problem but fail to keep up during heavy amounts of rain. And with rain and wet pavement representing 75% of all weather-related accidents in the U.S., I offer a newly innovated wheel well design that focuses on suppressing water spray while upholding the same standard for stopping all flying roadway debris like rocks, dirt, and mud. My engineering goal is to build a viable product to reduce the amount of water sprayed off a semi’s tires. To ensure a quality-working product, I have associated myself with an engineering design and development instructor, a mechanical engineer and drafter, licensed truck drivers, and even a fluid dynamics technical expert at the Volvo headquarters in Sweden. As of right now, semis are required by the U.S. Government to have mud flaps, and it is my hope to have the government implement my design as a requirement for all semis.