01) DigiClips Media Search Engine Bob Shapiro (DigiClips) bobshapiro40@gmail.com Problem we need solved: We are working on developing a Media Search Engine to process media data to create media results reports software for Television, Radio, Newspapers, Magazines, Social Media, Blogs, and Web Streaming Media to not miss data. To help people become media aware and media informed with Media Analytics. Work on Administration to keep track of computers and software hardware on systems with security hardware problems and error messages data missed with Administration Analytics. DigiClips Summary DigiClips Capstone DigiClips Order of Priorities DigiClips Project DigiClips overview
02) AI Recommendation Engine David Hughes (UNC and Kinetik Simulation) david.hughes@kinetiksimulation.com Develop an AI recommendation engine using AWS Comprehend service. We will use 100 manually developed use cases to train the AWS Comprehend engine. Kinetik overview vid Project overview vid
03) Machine learning to control UNC Wave Maker Rich McLaughlin (UNC Fluids Lab, UNC Math) rmm@email.unc.edu Builds on a former 523 class success in improving the unc wave maker interface software. Specificallly we want to try to understand how existing feedback to control the 30 piston engine wave engine can be improved using machine learning, with the goal of providing better performance. Former Wave Maker project Former proposal (23) News about Wave Maker
04) MySmile Volunteer Management System Sommer Wisher (Wake Smiles) sommer@wakesmiles.org We have worked with students in this course for 2 semesters that have successfully created an amazing volunteer management application. From this, volunteers for our nonprofit can sign in and register for shifts. We have an admin side that can log and track hours. Now that we have been working with it, we see some room for changes and improvements. We would love for ALL of our volunteers types to utilize this platform (currently only pre-dental students can use it). We would love for the information that admin can retrieve to be different and more streamlined, and for the new group to work out kink/bugs that we notice as we utilize the program. Wake Smiles vid Developer Doc
05) CyberLand Eyup Sabir Erbicer (UNC Dept. of Human Development and Family Sciences) erbicer@ad.unc.edu The aim of this proposal is to develop a mobile application that will increase adolescents’ awareness and prevention skills to cope with cyberbullying perpetration behaviors. Project vid
06) Kalman Filters for Apple Watches Dr. Andrew Abumoussa (UNC Neurosurgery) andrew@med.unc.edu (1) I have built an iOS app capable of performing Neuro navigation in real time. I would like to improve the tracking metrics of the phone by streaming acceleration data from the watch to the phone. The phone can then update position of a tracked object via a Kalman filter. This project will need a team working on Kalman filters - requires team have access to Macs + iPhones or IPad Pros and I’ll provide the team w an Apple Watch Series 8 for development. Neuro navigation vid Neuro navigation Project clent vid
07) Generative AI to Train Semantic Segmentation of Brain CT and MRIs Dr. Andrew Abumoussa (UNC Neurosurgery) andrew@med.unc.edu (2) Semantic segmentation allows for the automatic labeling of anatomic and pathological findings for both brain CTs and MRIs. For this project, I would like to work with a team to improve the accuracy of a semantic segmentation algorithm by using a GAN (generative adversarial network) to drive the training. A prior course’s team built a semantic segmentation algorithm capable of segmenting brain CTs with an accuracy of 98.7% and I would love to mentor a team to build and train a GAN to achieve an even higher accuracy without supervised learning. Former 523 Team Former proposal (9)
08) FANzPLAY...The Game within a GAME! Taylor Walls (Former UNC student, and CEO FanzPlay) htaylorwalls@gmail.com IF YOU COULD PLAY AGAINST THE FANS YOU LOVE TO HATE, WOULD YOU PLAY? FANzPLAY...The Game within A GAME! FANzPLAY is an early start-up sports/entertainment app that allows fans of the biggest rivalries in sports to play against each other, LIVE on Game Day to earn their own Bragging Rights and win gifts/prizes from our sponsors! FANzPlay App is Taking the Game to the fans, creating an exciting camaraderie from arenas to our living spaces and “Changing the way WE PLAY! This is a former 523 project, and we would like to continue its development. I will locate the previous team's work on the backend of the App which could be applied and work. Once that is reviewed, I would like for the team to create the UI/Front End. I do have sketches and files of the design and some marketing materials to give depth to the visual(s) with the teams input/direction of what is best. The video is only a concept and some of the features would be upgraded or deleted as discussed with the team. Fanz Play Walls vid
09) IPS Fidelity Project Penny Liles (UNC Dept of Psychiatry CECMH) penny_liles@med.unc.edu We want to create an online portal for behavioral health providers to input or upload data when a group of Institute and DHHS staff evaluate their team. We use this data to rate items based on a fidelity scale. Teams currently submit a spreadsheet with a lot of data and we make calculations on the data. The online portal would allow the evaluators to see the data in one spot and make calculations for ratings. We also use a consensus spreadsheet for ratings which we would like to add to the online portal (if feasible during the semester). Long-term we would like to add the report template to the portal. This project is a continuation of a Spring 2023 project in 523 where students turned an Excel spreadsheet (see attachment) we use for evaluating mental health providers into a website for providers to enter data. The students mostly completed the website and the next step is to format the website (visually), then start adding formulas so we can make calculations for ratings based on the evaluator spreadsheet (attached). Liles vid XLS fidelity XLS evaluator
10) Sustainablity Tracker Talpha Harris (CEO, Sustainible) talpha@sustainible.io High Level The Sustainability Tracker is a mobile-app that is for those who are interested in monitoring their environmental practices and contribution to sustainability. The app will monitor daily sustainability practices and make estimates such as personal trash output, food waste, and carbon footprint - based on input data. It will combine input data with daily monitoring and weekly questions, in order to generate daily recommendations via Azure AI, meant to increase personal sustainability scores. This will help address the problem of climate change and food shortages, and more - applicable to SDG goals. Technical Azure AI generates recommended outputs based on inputed data from prompts and api information. The input data and recommended output data is stored in a database so that an api can access the data then create read-only or read-write access files. Using node express, rest apis generate data points for external applications to use via json, xml. Potential API inputs -- Carbon Calculator: https://developer.mastercard.com/carbon-calculator/documentation -- Plaid: https://plaid.com/products/transactions/ -- Mileage tracker: https://www.microsoft.com/en-us/maps/bing-maps/distance-matrix Project vid
withdrawn 11) Medical Conference App Dr. Maria E. Ferris (UNC School of Medicine) mariaeferris@gmail.com Title: J.B. Graham Research Conference App. Objective: To build an app for the UNC Medical School J.B. Graham Research Conference. Methods: This is an annual conference and this year it will take place on November 13, 2023 at the Bill Roper Building (several floors and auditoriums). The app will: 1. guide the attendees about the day's schedule of activities and physical location of lectures and posters on the day of the conference. 2. host about 100 abstracts (PDF Files) with their corresponding posters (Usually power point files), about one month ahead of the conference (October 2023) 3. orient attendees of the physical location of posters across the building, for the day of the conference (November 13, 2023) 4. provide a platform (excel?) for a) faculty judges to virtually rate the abstracts and their respective posters and b) to identify the best posters that receive awards, with what we hope will be real-time reporting (November 1st to the 13th, 2023) 5. evaluate the conference format, location, activities and suggest new ideas for next year (November the 13th, to the 30th, 2023) 6. if possible, allow interaction between the attendees to leave comments, praises or even suggest ideas on how to improve the projects they read about. Alternatively, would we need to have a social media account to link year-round comments? (October-December, 2013) 7. suggestions are welcomed Note: the client knows the Oct-Nov "deadlines" and pie-in-the-sky but says the app will be helpful next fall and beyond
12) Shift Overflow Alyssa Lytle (UNC Comp Sci) abyrnes1@cs.unc.edu We have an existing scheduling software that some students of Kris Jordan wrote a while back (here). This site is used to manage the work hours and preferences of the dozens of LAs needed to teach COMP 110. However, it hasn't been maintained, and we'd love to have a team clean up and revitalize the site! Essentially, it's main purpose is to work so that it takes in a schedule of shifts from a manager, employee availability, and auto-schedules them to work those shifts. Project vid
13) DreamScape Gavry Eshet (CareYaYa Health Tech) gavry@careyaya.org (1) DreamScape is an avant-garde initiative aimed at visually reconstructing dreams by harnessing the combined power of EEG waves, artificial intelligence, and machine learning diffusion models. Derived from the intricate patterns of our brainwaves during sleep, DreamScape strives to make the intangible realm of dreams tangible, providing a glimpse into the subconscious mind. DreamScape will be created using React Native, as it will need to connect to a Muse headset via Bluetooth and transmit readings to a server. The product will display generated images, a textual interpretation, and graphical representations of different psychological factors (mood, stress, etc.). While DreamScape's groundbreaking approach intrigues a broad spectrum of users, the primary audience includes neuroscientists, psychologists, dream enthusiasts, and therapy patients. By offering a visual rendition of dreams: -- Researchers can delve deeper into dream analysis, contributing to the broader understanding of human cognition and subconscious processes. -- Therapists can employ DreamScape as an innovative tool for aiding psychoanalytical sessions, interpreting dream symbolism more effectively. -- For the general public, it offers a novel way to understand, remember, and share their dreams, fostering a deeper connection with their subconscious mind. CareYaya is a client in past 523 projects.
14) Generative Art Therapy Games Gavry Eshet (CareYaYa Health Tech) gavry@careyaya.org (2) The "Generative Art Therapy Games" project is an innovative confluence of generative AI art and therapeutic practices tailored for hospice care environments. Recognizing the profound emotional and psychological impact hospice settings can have on individuals, this project aims to offer an engaging and cathartic medium for self-expression and healing. By gamifying the art therapy process, it seeks to instill joy, intrigue, and comfort in its users during a challenging phase of their lives. The interface is accessed through a React web app, often on a tablet such as an iPad. This will allow for hospice patients and family members to make low-fidelity sketches and transform them into high-definition, realistic artwork. Ideally, this would empower non-artistic family members to engage in and lead art therapy for their loved ones. The project predominantly serves hospice patients, offering them a unique and engaging method to communicate emotions, recall memories, or merely find distraction and pleasure. CareYaya is a client in past 523 projects.
15) CollectiveGood Data Collection Sophia Fort (Carolina Population Center, CollectGoodMD) srfort@live.unc.edu CollectiveGood is creating an application, CollectiveGood MD, that will serve as a way for physicians around the world to access a quality second-opinion network right from their phone or laptop. There is an emerging body of literature in medicine showing that diagnostic opinions generated from collective intelligence of all levels of medical professionals (registered nurses, physician’s assistants, etc.) consistently outperform diagnoses from individual physicians. The strength of collective intelligence is relatively easy to harness in urban areas where physicians have connections to a large network of doctors from all areas of expertise. In rural areas, building a community of fellow physicians can prove exceedingly difficult both due to physical distance and other factors. Using CollectiveGood MD, physicians from virtually everywhere will be able to utilize collective intelligence for diagnostic support and access a crowd of fellow clinicians who have proven expertise in specific clinical domains. Users are also able to contribute their own knowledge and receive compensation for their evaluation of clinical cases, adding to the collective body of knowledge on the app. A major component of our application is the “Skill Test” feature, which is a learning feature where physicians can improve their diagnostic abilities through practice with clinical case vignettes from educational resources. In this feature, physicians are shown a patient case and are then asked to give three possible diagnoses for the case, as well as an explanation for each diagnosis and a confidence level to show how sure they are in each diagnosis. For this component of our application to run, we need to collect a large number of patient cases from a variety of sources. Collecting these cases can be time-consuming and inefficient -— we have to first locate appropriate sources from which we can collect cases, and then figure out how to abstract the cases from each source. Some are written directly into the HTML of the webpage; others are in PDF format or another type of document. So far, we have been entering these cases into our own NoSQL database by hand, but we would like to create a platform where we can outsource this data entry. We would like to create a platform where users can access previously identified raw patient data (could be in a table within a PDF document, a paragraph explaining patient symptoms, etc.), then work to enter the patient data in the correct format for use in our main application. This would be a web application. Users would log on, see their assigned data sources, then enter a platform where they see the raw data on one side of the screen and a form to organize the data on the other. We would want a way to validate that this data is entered correctly, so we would systematically assign the same patient records to multiple users. We would also want to establish a payment mechanism where individuals are paid for the cases they submit. This payment system would be dependent on the accuracy of their submission (determined by cross-validation) and other factors. Another way to cross-validate user entry would be to use an LLM to do a first pass at entering data then have our human users correct the entry by the LLM. Our priority at this time is to build the basic structure of the application. The payment and LLM components would be great to have after this semester but are not fundamental to what we would want from our project group. Client vid PPT on CollectGood
16) Swayambhu Stories: Visualizing Religion and Heritage in Nepal Dr. Lauren Leve (UNC Dept. of Religious Studies) leve@email.unc.edu This project involves documenting and preserving Buddhist cultural heritage in Nepal, using annotated VR models. Your coach this semester (Sam Shi) was a member of the original project team. Thus you have good help for this proposed continuation. Client vid Former COMP 523 Project Project description Supporting material
17) Buckner (late submission)