A Decision Guidance System for Precision Agriculture Management in a Nonstationary Environment using Advanced Machine Learning Algorithms
Team members: Viviana Maggioni from Department of Civil, Environmental, and Infrastructure Engineering; Alexander Brodsky from Computer Science, Yuan Xue and Paul Houser from Department of Geography and Geoinformation Science
Water, energy and nutrient resources are the primary determinants of productivity in agricultural ecosystems. In a nonstationary environment, agricultural practices based on historical experience may fail. New approaches of dynamic monitoring and adaptation of agricultural management to changing resource levels will be needed to achieve socio-ecosystem sustainability. To this end, we propose an advanced stochastic system for adaptive precision agriculture in a nonstationary environment that not only monitors but also forecasts agricultural resource conditions to guide optimal management as well as to make actionable recommendations. Our goal is to develop a decision guidance system (DGS) to empower agriculturalists to adapt to changing environment, climate, and practice, impacting irrigation, water efficiency, planting, harvesting, pruning, weed control, fertilizers, pesticides, herbicides, crop production and quality, and disease control. Our research will progress agricultural decision support from classic “almanac” methodologies to project crop conditions and resource management dynamically, using computational Earth-system modeling techniques and machine learning approaches. The system will move far beyond traditional empirical crop system models to a dynamic, physical atmosphere-soil-vegetation land models that resolve the diurnal cycle, including complete biogeophysics and dynamic phenology, and predicts optimal water, energy, carbon, and nutrient resource use. The proposed project is particularly relevant to IDIA’s research goals, as it focuses on critical emerging areas of research (i.e., nonstationary theory), is highly interdisciplinary (i.e., geography, agriculture, remote sensing, engineering, and computer science), is a combination of both sound biogeophysics (i.e., model calibration phase) and advanced machine learning techniques (i.e., model projection phase), and has great potential not only to advance but also transform precision agriculture integrative modeling and prediction tools for nonstationary resource optimization and productivity. Therefore, the proposed DGS can be a useful tool to support and provide actionable recommendations to diverse stakeholders in the areas such as water availability (irrigation scheduling, sustainability, drainage water management and control, alternative water sources etc.), quality (controlling soil erosion, transport and fate of sediments and contaminants etc.), and conservation (selecting placing and combining practices, new designs, equipment and materials).
Smart Music Intervention Program for Older Adults with Cognitive Impairment: A Protocol Development
Team members: Emily S. Ihara, Department of Social Work; Parth Pathak, Department of Computer Science, Y. Alicia Hong, Department of Health Policy & Administration; Huzefa Rangwala, Department of Computer Science; Cathy Tompkins, College of Health & Human Services/Department of Social Work; Megumi Inoue, Department of Social Work
The goal of this project is to develop an easily accessible, automatic, personalized digital music intervention program for older adults living with cognitive impairment. The team will build upon an evidence-based, nonpharmacological personalized music listening intervention that has been shown to decrease negative psychological and behavioral symptoms for individuals living with cognitive impairment and gained traction in nursing homes and long-term care organizations nationally and internationally. Given the benefits of personalized music for individuals living with dementia, development of this type of technology will provide a mechanism for the timely delivery of the appropriate music intervention through a wearable device that collects physiological measures through sensors. Through the prototype development process, we will triangulate the physiological, observational, and self-reported effects of personalized music for individuals living with dementia which will further inform how to further digitize the intervention, allowing for the scaling up of the intervention in a large randomized clinical trial. Taking this affordable, non-pharmacological intervention to the next level provides the opportunity to reach many more individuals and organizations and improve the lived experiences for those living with dementia.
Understanding the Impact of Misinformation on Palliative Care Demand Using Machine Learning and Qualitative Methods
Team members: Megumi Inoue from Social Work, College of Health and Human Services; Mahdi Hashemi from Information Sciences and Technology at College of Engineering and Computing; Naoru Koizumi and Rajendra Kulkarni from Schar School of Policy and Government; Denise Mohess from Inova Fairfax Hospital; Matthew Kestenbaum from Capital Caring Health
Palliative care is one of the fastest-growing medical specialties in health care. Its value has become particularly evident during the COVID-19 pandemic. At the same time, palliative care faces various challenges including misconceptions among the general public, a lack of awareness of its benefits, and limited and sporadic access and coverage by Medicare, and a lack of proper regulatory and accrediting bodies which result in an insufficient number of high-quality specialists. This project explores a new methodological framework to identify reasons for and factors associated with the underutilization of palliative care in the United States. The methodological framework takes advantage of traditional qualitative research approaches of semi-structured interviews and focus groups coupled with automatic mining of online information sources, including Twitter and Google Alert. The main outcomes of this research are preliminary evidence showing: i) various types of misinformation in relation to palliative care; and (ii) that the analytical framework developed under the project can provide new insights to inform future strategies for palliative care. By achieving these outcomes, this project can help healthcare organizations to strategize their information dissemination in palliative care and work effectively with patients and their families.
LegisSciT: A Cooperative Longitudinal Data Tracking Platform for the International Study of Legislative Science Advice
Team members: Karen Akerlof from Department of Environmental Science & Policy; Dieter Pfoser from Department of Geography and Geoinformation Science; Erica Goldman, The National Council for Science and the Environment (NCSE)
Karen Akerlof and Dieter Pfoser of George Mason University’s departments of Environmental Science & Policy and Geography & Geoinformation Science are partnering with Erica Goldman at the Global Council for Science and the Environment in a new Institute for Digital Innovation seed grant titled “LegisSciT: A Cooperative Longitudinal Data Tracking Platform for the International Study of Legislative Science Advice.” The interdisciplinary project will generate new data and methods to answer fundamental questions about the impact of science on society and transform the emergent field of legislative science advice. The proposed research presents a novel methodological approach for the study of individual, group, and institutional factors that are hypothesized to influence the way in which policymakers use science. The project complements an ongoing joint effort between the university and GCSE.
Advancing Workplace Accessibility by Improving Neurodiverse Individuals’ Executive Function via Wearable Technology
Team members: Vivian Genaro Motti from Information Sciences and Technology at School of Computing; Sarah Wittman and Richard Klimoski from School of Business; Heidi Graff from SourceAmerica; April Pinch-Keeler from MVLE
This project investigates how assistive wearable technology can promote self-efficacy and improve executive function for neurodiverse employees executing manual activities and seeking to maintain autonomy and performance. The innovative wearable technology proposed, via a smartwatch application powered with AI models will monitor human performance—recognizing work activities and intervening through personalized, strategic, just-in-time prompts. Partnering with industry collaborators from SourceAmerica and MVLE, a multi-disciplinary team of engineering, business and organizational psychology, and special education experts aims to improve neurodiverse adults’ workplace accessibility.
The project adopts a user-centric inclusive design approach to elicit the potential opportunities of a wearable technology to address the challenges, needs, and desires of neurodiverse and neurotypical workers and their employees. Via a controlled laboratory experiment the team will generate recognition models from wearable sensor data (inertial measurement units). Then, in a controlled experimental design pilot study (3x3x2 study), the team will evaluate the application’s impacts on subjective (self-efficacy, job satisfaction, independence) and objective (executive function, productivity, accuracy) dependent work measures for neurotypical and neurodiverse individuals.
Besides integrating neurodiverse adults in the research agenda, this project will lead to more inclusive work practices by introducing an innovative wearable technology that will promote workplace access for currently under-employed neurodiverse adults.