JTJordan Tschida

Jordan Tschida

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Jordan Tschida has a total of 28 co-authors. Here are the top co-authors by number of shared publications:

  • Troy McDaniel

    Arizona State University

    4 shared publications
  • Mayanka Chandrashekar

    Computational Sciences and Engineering Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee, United States of America

    3 shared publications
  • Daniel Santel

    Division of Biomedical Informatics, Department of Pediatrics, Cincinnati Children’s Hospital Medical Center, University of Cincinnati College of Medicine , Cincinnati, OH, 45229

    2 shared publications
  • Heidi A Hanson

    Advanced Computing for Health Sciences, Computational Sciences and Engineering Division, Oak Ridge National Laboratory , Oak Ridge, TN, 37830

    2 shared publications
  • Ian Goethert

    Information Technology Services Division, Oak Ridge National Laboratory , Oak Ridge, TN, 37830

    2 shared publications
  • John Pestian

    Division of Biomedical Informatics, Department of Pediatrics, Cincinnati Children’s Hospital Medical Center, University of Cincinnati College of Medicine , Cincinnati, OH, 45229

    2 shared publications

Latest Publications

  • Evolving language of pediatric anxiety in electronic health records

    by Jordan Tschida, Mayanka Chandrashekar, Heidi A Hanson, Ian Goethert, Daniel Santel, John Pestian, Jeffery R Strawn, Tracy Glauser, Anuj J Kapadia, Greeshma A Agasthya

    Objectives: This study aimed to identify and quantify semantic drift (ie, the change in semantic meaning over time) within expert-defined anxiety-related (AR) terminology and compare it to common electronic health record (EHR) vocabulary across longitudinal pediatric clinical notes. Materials and Methods: A corpus of pediatric clinical notes from 2009 to 2022 was analyzed using computational methods. Semantic drift for each term was quantified using cosine similarity between annual temporal word embeddings. Contextual meaning was examined through changes in nearest neighbors across years. The Laws of Semantic Change were applied to assess the influence of word frequency and polysemy. Vocabulary terms were categorized as AR or common EHR. Results: 98% of AR terminology maintained a cosine similarity between 0.00 and 0.50, indicating moderate semantic stability, whereas 90% of common EHR terms remained between 0.00 and 0.25, showing greater contextual stability overall. Frequent terms exhibited minimal change (Frequency Coefficient = 0.04), whereas highly polysemous or abbreviated terms showed less stability (Polysemy Coefficient = 0.630). AR terminology drifted more slowly than general EHR vocabulary (Type Coefficient = −0.179), further supported by significant year–type interactions (Coef = −0.09 to −0.523). Discussion: Although anxiety-related terminology demonstrates slower semantic drift than general EHR vocabulary, subtle contextual shifts still occur that may affect downstream interpretability and retrieval in automated systems. Conclusion: Continuous linguistic monitoring and adaptive modeling are essential to maintain semantic fidelity and ensure the long-term reliability of clinical decision support systems as healthcare documentation evolves.

  • Open Access

    Mitigating Algorithmic Bias in Cancer Site Classification Models

    by Abhishek Shivanna, Adam Spannaus, Jordan Tschida, John Gounley, Patrycja Krawczuk, Heidi Hanson

    PURPOSE: Integrating artificial intelligence in cancer diagnostics has improved tumor classification beyond rule-based systems. Despite these advancements, these models may still encode demographic biases. We conducted a large-scale, applied bias-probing study of a deep learning–based cancer site classifier to quantify race information encoded in document embeddings. We then evaluated how performance changes when race-correlated embedding dimensions are removed in a post-training sensitivity analysis. METHODS: The cancer site classifier was trained using 3.5 million electronic cancer pathology reports from six of the National Cancer Institute's SEER registries. We trained a hierarchical self-attention network to generate 400-dimensional document embeddings. These embeddings were used to train two downstream, gradient-boosted decision tree classifiers: one to classify the cancer sites and another to predict racial categories. We identified overlapping features by intersecting the top 50 feature-importance rankings from the site and race models and computed their cumulative feature importance in each model. As a post hoc sensitivity analysis, we progressively pruned these overlapping dimensions, retrained the site model, and compared overall macro-F1 and accuracy, race-stratified macro-F1, and group fairness metrics on the basis of demographic parity and equalized odds before and after pruning. RESULTS: The analysis revealed minimal feature overlap between the cancer site and race prediction models, and the cumulative importance scores indicated a negligible influence of racial information on clinical predictions. Post-training pruning of overlapping features did not compromise the models' diagnostic accuracy, with a 0.07% loss in accuracy. CONCLUSION: Our findings demonstrate that HiSAN-generated embeddings from SEER data can be used effectively in cancer site classification without significant demographic bias influencing the outcomes. Post-training pruning therefore functions as a practical audit and sensitivity check.


  • Evaluating algorithmic bias on biomarker classification of breast cancer pathology reports

    by Jordan Tschida, Mayanka Chandrashekar, Alina Peluso, Zachary Fox, Patrycja Krawczuk, Dakota Murdock, Xiao-Cheng Wu, John Gounley, Heidi A Hanson

    Objectives: This work evaluated algorithmic bias in biomarkers classification using electronic pathology reports from female breast cancer cases. Bias was assessed across 5 subgroups: cancer registry, race, Hispanic ethnicity, age at diagnosis, and socioeconomic status. Materials and Methods: We utilized 594 875 electronic pathology reports from 178 121 tumors diagnosed in Kentucky, Louisiana, New Jersey, New Mexico, Seattle, and Utah to train 2 deep-learning algorithms to classify breast cancer patients using their biomarkers test results. We used balanced error rate (BER), demographic parity (DP), equalized odds (EOD), and equal opportunity (EOP) to assess bias. Results: We found differences in predictive accuracy between registries, with the highest accuracy in the registry that contributed the most data (Seattle Registry, BER ratios for all registries >1.25). BER showed no significant algorithmic bias in extracting biomarkers (estrogen receptor, progesterone receptor, human epidermal growth factor receptor 2) for race, Hispanic ethnicity, age at diagnosis, or socioeconomic subgroups (BER ratio <1.25). DP, EOD, and EOP all showed insignificant results. Discussion: We observed significant differences in BER by registry, but no significant bias using the DP, EOD, and EOP metrics for socio-demographic or racial categories. This highlights the importance of employing a diverse set of metrics for a comprehensive evaluation of model fairness. Conclusion: A thorough evaluation of algorithmic biases that may affect equality in clinical care is a critical step before deploying algorithms in the real world. We found little evidence of algorithmic bias in our biomarker classification tool. Artificial intelligence tools to expedite information extraction from clinical records could accelerate clinical trial matching and improve care.


  • Characteristics of social robots: A framework for development

    by Jordan Tschida, Katina Michael, Troy McDaniel

    Social robots hold great promise for assisting both older adults with dementia and those in good health. Yet, little is known about how people want to interact with these robots. This article introduces a framework for the key features social robots should have, based on insights from a four-month study of how seniors engage with robots through conversation and interaction. We have broken these features down into principles of robot behavior, verbal communication, and nonverbal communication, while also considering factors that affect a person’s likelihood to adopt such technology. Our findings offer a starting point for refining social robots and invite further exploration into developing features that truly resonate with users.


  • Open Access

    Temporal Drift in the Semantic Meaning of Pediatric Anxiety Terms in Electronic Healthcare Records

    by Jordan Tschida, Mayanka Chandrashekar, Heidi A. Hanson, Ian Goethert, Surbhi Bhatnagar, Daniel Santel, John Pestian, Jeffery R. Strawn, Tracy Glauser, Anuj J. Kapadia, et al.

    : Objective: To identify and measure semantic drift (i.e., the change in semantic meaning over time) in expert-provided anxiety-related (AR) terminology and compare it to other common electronic health record (EHR) vocabulary in longitudinal clinical notes. Methods: Computational methods were used to investigate semantic drift in a pediatric clinical note corpus from 2009 to 2022. First, we measured the semantic drift of a word using the similarity of temporal word embeddings. Second, we analyzed how a word’s contextual meaning evolved over successive years by examining its nearest neighbors. Third, we investigated the Laws of Semantic Change to measure frequency and polysemy. Words were categorized as AR or common EHR vocabulary. Results: 98% of the AR terminology maintained a cosine similarity score of 0.00 – 0.50; at least 90% of common EHR vocabulary maintained a cosine similarity score of 0.00 – 0.25. Laws of Semantic Change indicated that frequently occurring vocabulary words remained contextually stable (Frequency Coefficient = 0.04); however, words with multiple meanings, such as abbreviations, did not show the same stability (Polysemy Coefficient = 0.630). The semantic change over time within the AR terminology was slower on average than the semantic change within the common EHR vocabulary (Type Coefficient = -0.179); this was further validated by interacting the year and Type (Coef = -0.09 – -0.523). Conclusions: The semantic meaning of anxiety terms remains stable within our dataset, indicating slower overall semantic drift compared to common EHR vocabulary. However, failure to capture nuanced changes may impact the accuracy and reliability of clinical decision support systems over time.

  • Exploring Social Robots for Healthy Older Adults: Aging With Companionship

    by Jordan Tschida, Katina Michael, Troy McDaniel

    Loneliness and social isolation are prevalent among older adults and are associated with adverse health outcomes. Social robots offer a novel approach to addressing these issues by providing companionship and social support. This study examines the preferences of older adults when conversing with social robots that use verbal and nonverbal communications. The methodology of this study incorporated both quantitative and qualitative approaches. Sixteen older adults residing in an independent living facility participated in a 4-week study, during which they were observed interacting with a social robot in weekly sessions. The study employed a Wizard-of-Oz experimental design to investigate verbal and nonverbal communication levels. Data was collected in three phases beginning with the human-to-robot conversation observation immediately followed by a post-interaction survey, open-ended interviews, and finally a post-experience survey. Participants reported positive experiences with the robot, including companionship, enjoyment, and emotional support. The robot’s ability to remember details about participants and engage in responsive conversation was highly valued. All participants desired the robot to have more verbal and nonverbal communication skills. The preliminary findings suggest that social robots have the potential to mitigate loneliness and enhance social connectedness among older adults. Further research with more diverse samples is warranted to validate these findings and explore long-term effects. Addressing ethical considerations will be crucial to maximize the benefits of social robots in promoting the well-being of aging populations.


  • Next Steps for Social Robotics in an Aging World

    by Jordan Miller, Michael Bernstein, Troy McDaniel

    Scientists and engineers are exploring technological solutions to place inside people’s homes to solve challenges associated with rising healthcare needs and an aging population. This trend in technology is known informally as smart living. The technologies being investigated may hold a promising future for the elderly population, allowing them to continue to live inside their home while aging. However, when novel devices are introduced to support aging in place, designs often fail to consider the struggles older individuals face in their everyday lives. Excluding such considerations means continuing to develop technologies that do not truly serve the needs of older adults. Earth is projected to be home to 9.3 billion people by 2050, 2 billion of which are likely to be 65 or older [1] . Therefore, there is a pronounced need for a human-centered approach to appropriately address the needs, wants, opinions, behaviors, and psychosocial aspects of older adults desiring to age in place.