Our NSF project pitch was successful! After creating a multidisciplinary team with Atmospheric Sciences, Environmental Sciences, School of Natural Resources, Organicforecasting LLC., and other small businesses, we submitted a proposal to build a system for the nation to use related to predicting weather. The system would include weather data and provide a forecast between 15-35 days, 35-42 days, and yet another set that is vastly ignored from 42-84 days. Being able to accurately predict these forecasts has not been done before. We utilize a special blend of datasets, which are not being used elsewhere, and plan on expanding this methodology using machine learning to improve verification rates more.

After winning a pitch and securing a grant from the Entrepreneurship Alliance program, hosted by the Center for Entrepreneurship and Innovation (CEI) at the University of Missouri, we were tasked with developing an application that worked on all major platforms (e.g. Android, iOS, and web) that would search and scrap data from the web and popular social media platforms (e.g. Instagram) and display local specials and discounts at restaurants, bars, and various local businesses. Users could see these discounts and then use our app to find the business, basic business information such as hours, and display the coupon/deal at the business to get the discount.

A multidisciplinary team was created to develop a mobile application for the State of Missouri Historical Society. We used a MEAN stack and popular frameworks like Angular and Ionic to develop a cross-platform application. The app would allow the State Historical Society of Missouri to enter in all events, users could see those events in real-time, save events, add events to calendar, see others who plan on attending events, and can provide real-time feedback about the event. Essentially this application was a social media event application for all of the State of Missouri events and residents. After being successful, this collaboration opened up many doors working with various government agencies and private organizations throughout the midwest.

Word quickly spread about the successful projects we designed and developed. Many projects were being presented to Streamline. The Industrial and Manufacturing Systems Engineering (IMSE) chair reached out and wanted a website developed that would keep track of all IMSE alumni as a private social media site. A website and non-relational database were created. In addition, a Python web scraper was built to pull information, such as LinkedIn profile information, and store that information to the secure database, so people can stay up-to-date with other colleagues, all happening in real-time automatically. IMSE staff and faculty would upload a list of currently graduating students, the scraper would run and pull corresponding information at scheduled intervals, alumni could submit an application to be considered for the hall of fame on the website, and once accepted, hall of fame alumni would be given special privileges. More functionality for the social media site is in discussion and will be implemented in phase 2.

The occurrence of severe weather is an annual problem for much of the United States and North America from March through July. With the increased interest in sub-seasonal weather forecasting, there have been attempts to anticipate the occurrence of anomalous weather on the time scale of one to four weeks including the occurrence of severe weather. Previous research has shown that teleconnection indexes, associated with long period Rossby Wave activity, or persistent large-scale flow regimes have been useful tools in this endeavor. Here, abrupt changes over a 24 – 72 hour period in the Southern Oscillation Index (SOI) time series will be used to demonstrate that these changes can be associated with the possible occurrence of major severe weather event-days (e.g. 20 or more tornadoes, 155 or more wind events > 25.9 meters per second, or 135 hail events larger than 25.4 mm) over the US one to three weeks in advance, especially during the severe weather season. The severe weather events obtained from the archive at the Storm Prediction Center (SPC) from 1991 through 2020 were used. The results here demonstrate that more than seven in ten major severe weather occurrences were associated with abrupt changes in the SOI when using a strict test of the predictability.

A 3-year proposal was submitted to the NSF for 1.2 million in order for us to design and build a multidisciplinary team of Soil Microbial Ecologists, Soil Health Experts, Soil and Environmental Sciences, Plant Science & Technology, IT Professionals, Computer Scientists, Machine Learning Engineers, and Software Engineers. This team would collaborate to build a real-time intensive cyber-infrastructure support system with data fusion from soil-based sensors with remote sensing data to provide high-resolution spatial coverage of soil properties. This real-time cyber system will be used for secure collection, storage, and sharing of massive amounts of data by many wireless sensors. A dashboard was created to innovative data analytics and visualization (i.e. Geovis) tools to enable quantitative data from soil sensors to be represented and interpreted over space and time. A phase 2 application will be submitted in order to analyze the data and use machine learning and deep learning to predict soil quality and best areas for growth on specific plant species.

Many projects were being brought to Streamline after all the good reviews from all the good work we had completed. The Director of the Equine Endowed Program at the University of Missouri reached out to us and needed work completed. They had a 20+ year dataset that needed analyzing for patterns. The final result was to use the data collected from sensors to predict ataxia and lameness in horses. They would attach sensors to each leg, the head, and the pelvis of the horse. The horse would perform tasks, and we would use that data to determine if the horse was ataxic or lame. After many rounds of collaboration with many multidisciplinary team members, we were able to reach high accuracy (i.e. precision and recall), and we were able to have a low false positive rate. False positives are important for this project because if a horse is predicted as positive, they need to go through an intensive procedure to determine if they have ataxia or lameness, which is very invasive for the horse and very expensive. Since we were working with the University, the research was published in scholarly articles and many students used this work as their graduate thesis.

Mizzou was sending too many emails to students and was flooding their inboxes with updates about events, seminars, news, course updates, etc. Students were overwhelmed and were not getting important emails about special events like large companies coming to campus (i.e. Google, Microsoft, etc.). Students were reporting not checking their emails because of the mass amount of emails they received each day. The college of engineering wanted an application where they can post events to students, faculty, and the public. This project will have several stages, the first stage will consist of having events show up to the correct people with push notifications, getting feedback to know who plans to come to the event, allowing the user to give feedback about the event, having the user share the event with their friends, and much more for future stages.

Working with the Digital Biology Lab and Biochemistry, we wanted to create a one-stop shop for everything dealing with seed oil. Increasing seed oil content by plant breeding has resulted in trade-offs or penalties with respect to protein content, seed size, or seed set. The molecular basis for this impasse is mostly speculative. Use of current global profiling approaches to better understand both the metabolic consequences of higher oil and the basis for reduced yield must also deal with off-target genetic mutations (even in near-isogenic lines), ultimately confounding cause-effect interpretations. We propose a diverse, integrated strategy to study the consequences of higher lipid production by studying transgenic plants specifically engineered to produce higher seed oil.

A research community to collaborate on research projects and a tool for students and other public users to find various research projects and activities throughout the world. A collaboration network is generated as users add accounts. Each user has a private page and a public page. The private page is only available to users in the Yaffle system. The public page is viewable by the public and used as advertising for that user to connect with people around the world. A heat map of the research projects around the world is generated for global connections.

A dashboard created to help Mizzou manage their massive amounts of research data in real-time while allowing administrators to focus on analyzing the data rather than trying to find and manage the data. Mizzou has large amounts of research data including research expenditures, research proposals, grants received, grants submitted, who is working on which project, and the list goes on. This data is classified by different metrics like by shared credit, managing unit, etc. The dashboard organizes the data and displays results visually. Data breakdowns by university, college, department, and individual investigator are reported. This will help promote collaboration, minimize error, and recognize faculty members for helping grow innovative research at Mizzou's four universities.

In this project, several state-of-the-art deep learning models and their combinations have been applied to fish recognition in images, in particular 9 common species of fish in Missouri rivers. Four different data processsing and maching learning pipelines have been developed and extensive experiments have been conducted to evaluate their performances to form a working product for the Missouri Department of Conservation.

The Business School at the University of Missouri was impressed with what we could do, which brought them back for more. We decided to focus more on the transcripts and audio data combined, we generated our own features as well as state-of-the-art features, and developed a new system. This work focuses on determining sentiment from call center audio records, each containing a conversation between a sales representative and a customer. The sentiment of an audio record is considered positive if the conversation ended with an appointment and is negative otherwise. In this project, we developed a data processing and maching learning pipeline.

For this project, audio sentiment analysis was applied to call center conversations between a salesperson and a customer to predict if the customer would set up a sales meeting based upon their sentiment during the call. The two systems developed from this project are an end-to-end sentiment analysis pipeline for segmenting, performing feature extraction, and classifying conversational audio files using classical machine learning methods and a second end-to-end sentiment analysis pipeline that utilizes a deep Recurrent Neural Network to predict sentiment.

Measure Square, Inc. in Pasadena, CA had a massive dataset of floor plans for buildings, houses, and various structures. They were calculating the results by hand and wanted a way to automatically calculate different features using image processing techniques. They came to us and we decided we could develop the algorithms. We started out implementing image processing techniques but quickly went to machine learning. There we develop various algorithms and generated the results they were after.

The Missouri Department of Conservation (MDC) has massive amounts of imagery data for birds around Missouri. With that data, they try to keep track of wildlife and vegetation around the state. One large bird population that needs to be accounted for are Waterfowl. From the image, MDC would manually count the number of birds by hand. Instead, they reached out to us in order to develop a machine learning algorithm that would count the number of birds automatically. This application is powerful and has great potential for future work in different areas.

This project provided a machine learning pipeline that would take psychophysiological data and classify mod dysregulation using various algorithms. The best results obtained were approximately 90% accuracy when adding categories for the time of day. This project was used by the Psychology Department at the University of Missouri for one of their large grant proposals.

Our previous hybrid app was very successful and MDC liked the idea of having the application on any mobile operating system. Therefore, they wanted a new application for creel

We were collecting data for the Psychology projects for approximately a year and the system was solid. Now our Psychology collaborators wanted to use this data to predict whether the person was drinking alcohol or not. The system was continuously collect physiological data like heart rate, breathing rate, ECG, activity, and many more. We would use this data and analyze it in real-time to predict whether the user started to drink alcohol using machine and deep learning techniques. If we determined the usesr was drinking, we would prompt the user to complete a survey which would then allow us to better understand what causes people to drink.

Streamline Technology and the Missouri Department of Conservation's (MDC) relationship continued to build. In Missouri, Chronic Wasting Disease (CWD) is a serious problem and has a possibility of harming humans. MDC wanted an application where users could input their location, or have it automatically collected, input how many deer they saw with this illness, and have the data collected for processing. This would help Missouri keep track of this disease and help manage it to protect humans.

The Missouri Department of Conservation wanted to release an application to the public and they determined most users do not have Android based systems and they would need an application that would work on all mobile operating systems. Therefore, we developed this as a Hybrid Mobile Application where the app could be installed on any operating system. This app was used to collect user observations of animals they saw such as deer. The app would collect the user's location and how many animals they saw, almost like a hunting log. The log could be shared amoung users or kept privately. This app would help the conservation department to keep track of wildlife populations and help users determine the best spot to hunt.

The good word continued to spread and another psychology research lab receved a grant from NIMH and needed an application to collect user input, brain scans, and physiological data collected from Hexoskin. We developed this system and later we would analyze the data for them to predict stress and mood dysregulation in real-time.

The Missouri Department of Conservation was so impressed with our system because it was configured and customized to exactly what they wanted, we began working on the black bear data visualization and analysis project in parallel with the Shooting Range System.

We begin working on the shooting range data collection and visualization system. We develped an Android based mobile appication with an HTML5 web-based visualation tool. MDC was very satisfied and wanted other custom technology tools.