# Public health ![public health](/sites/g/files/omnuum6171/files/styles/hwp_1_1__720x720_scale/public/crcs/files/public_health_water_spicket.jpg?itok=wxV7n4-N) The focus of Harvard CRCS is to advance computer science research that serves the public interest by prioritizing topics in conservation and public health. CRCS researchers use computing tools and research methods in Artificial Intelligence to help augment decision making on a variety of topics related to health, social services, and wildlife conservation. Problems of prediction, identification, and planning help to leverage and optimize existing resources. Public health topics of current interest include mobile health delivery, tuberculosis medication adherence, COVID 19 spread, suicidality and metal health topics. **Project spotlight:** [**Boston Unemployment Map**](https://hermansaksono.github.io/boston_unemployment_vis/) [Action Boston for Community Development (ABCD)](https://bostonabcd.org/) and CRCS researchers worked together to develop a visualization tool to identify and show unemployment among racial and ethnic groups. With 60 years of community action and neighborhood engagement in Boston, ABCD is interested in enhancing their workforce development and support programs for job seekers. More specifically, they see that innovative technologies will allow them to make strategic decisions in allocating resources including outreach efforts. Using a user-centered and iterative design approach, the ABCD and CRCS decided to develop a [visualization tool that shows unemployment disparities in Boston at sub-neighborhood levels](https://hermansaksono.github.io/boston_unemployment_vis/). This tool allows them to see parts of neighborhoods in the city where there is disproportionate unemployment among Black and Latino women and men. While this tool provides the insights required by ABCD to make strategic decisions, the CRCS team pointed out that the tool is also emphasizing the social injustice issues that continue to linger in Boston. The unemployment disparities are more pervasive within the neighborhoods where Black and Latino households live. Therefore, without more targeted interventions, these employment disparities constrain Black and Brown people from achieving economic mobility. **Optimal resource allocation for mobile health clinics based on data-driven demand prediction** Mobile clinics are a viable solution to fight against both the COVID-19 pandemic and health disparities. This project aims at helping mobile clinics to intelligently allocate their resources using prediction and optimization approaches. In collaboration with researchers and practitioners at [The Family Van](http://www.familyvan.org/), a local non-profit mobile health clinic, and Harvard Medical School, CRCS researchers proposed an AI-based demand prediction method to help forecast the future demand of mobile clinics. Next steps in this collaboration include discussions about how the proposed demand prediction may be deployed to help the clinic’s daily operations. ## Publications on Equity Download 8 citations download- [BibTeX](/bibcite/export?pager_style=no_pager&number_of_items=8&sort_field=bibcite_year--desc&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B0%5D%5Btarget_id%5D=122862&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B1%5D%5Btarget_id%5D=126415&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B2%5D%5Btarget_id%5D=122765&taxonomy_filters%5Bfield_hwp_c_publicationsaiforsoc%5D&&&format=bibtex) - [EndNote X3 XML](/bibcite/export?pager_style=no_pager&number_of_items=8&sort_field=bibcite_year--desc&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B0%5D%5Btarget_id%5D=122862&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B1%5D%5Btarget_id%5D=126415&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B2%5D%5Btarget_id%5D=122765&taxonomy_filters%5Bfield_hwp_c_publicationsaiforsoc%5D&&&format=endnote8) - [EndNote 7 XML](/bibcite/export?pager_style=no_pager&number_of_items=8&sort_field=bibcite_year--desc&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B0%5D%5Btarget_id%5D=122862&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B1%5D%5Btarget_id%5D=126415&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B2%5D%5Btarget_id%5D=122765&taxonomy_filters%5Bfield_hwp_c_publicationsaiforsoc%5D&&&format=endnote7) - [Endnote tagged](/bibcite/export?pager_style=no_pager&number_of_items=8&sort_field=bibcite_year--desc&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B0%5D%5Btarget_id%5D=122862&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B1%5D%5Btarget_id%5D=126415&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B2%5D%5Btarget_id%5D=122765&taxonomy_filters%5Bfield_hwp_c_publicationsaiforsoc%5D&&&format=tagged) - [Marc](/bibcite/export?pager_style=no_pager&number_of_items=8&sort_field=bibcite_year--desc&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B0%5D%5Btarget_id%5D=122862&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B1%5D%5Btarget_id%5D=126415&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B2%5D%5Btarget_id%5D=122765&taxonomy_filters%5Bfield_hwp_c_publicationsaiforsoc%5D&&&format=marc) - [PubMedId](/bibcite/export?pager_style=no_pager&number_of_items=8&sort_field=bibcite_year--desc&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B0%5D%5Btarget_id%5D=122862&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B1%5D%5Btarget_id%5D=126415&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B2%5D%5Btarget_id%5D=122765&taxonomy_filters%5Bfield_hwp_c_publicationsaiforsoc%5D&&&format=pubmed_id) - [RIS](/bibcite/export?pager_style=no_pager&number_of_items=8&sort_field=bibcite_year--desc&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B0%5D%5Btarget_id%5D=122862&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B1%5D%5Btarget_id%5D=126415&taxonomy_filters%5Bfield_hwp_c_publications12345678%5D%5B2%5D%5Btarget_id%5D=122765&taxonomy_filters%5Bfield_hwp_c_publicationsaiforsoc%5D&&&format=ris) Gowda, S., Joshi, S., Zhang, H. & Ghassemi, M. [Pulling Up by the Causal Bootstraps: Causal Data Augmentation for Pre-training Debiasing](/publications/pulling-causal-bootstraps-causal-data-augmentation-pre-training-debiasing). Gowda, S., Joshi, S., Zhang, H. & Ghassemi, M. [Pulling Up by the Causal Bootstraps: Causal Data Augmentation for Pre-training Debiasing](/publications/pulling-causal-bootstraps-causal-data-augmentation-pre-training-debiasing). - add\_circle\_outline do\_not\_disturb\_on Abstract Machine learning models achieve state-of-the-art performance on many supervised learning tasks. However, prior evidence suggests that these models may learn to rely on shortcut biases or spurious correlations (intuitively, correlations that do not hold in... Singh, H., Joshi, S., Doshi-Velez, F. & Lakkaraju, H. [Learning under adversarial and interventional shifts](/publications/learning-under-adversarial-and-interventional-shifts). Singh, H., Joshi, S., Doshi-Velez, F. & Lakkaraju, H. [Learning under adversarial and interventional shifts](/publications/learning-under-adversarial-and-interventional-shifts). - add\_circle\_outline do\_not\_disturb\_on Abstract - [ descriptionPublisher's Version](https://arxiv.org/pdf/2103.15933.pdf) Machine learning models are often trained on data from one distribution and deployed on others. So it becomes important to design models that are robust to distribution shifts. Most of the existing work focuses on optimizing for either adversarial shifts... - [ descriptionPublisher's Version](https://arxiv.org/pdf/2103.15933.pdf) Parbhoo, S., Joshi, S. & Doshi-Velez, F. [Generalizing Off-Policy Evaluation From a Causal Perspective For Sequential Decision-Making](/publications/generalizing-policy-evaluation-causal-perspective-sequential-decision-making). Parbhoo, S., Joshi, S. & Doshi-Velez, F. [Generalizing Off-Policy Evaluation From a Causal Perspective For Sequential Decision-Making](/publications/generalizing-policy-evaluation-causal-perspective-sequential-decision-making). - add\_circle\_outline do\_not\_disturb\_on Abstract - [ descriptionPublisher's Version](https://arxiv.org/abs/2201.08262) - [ picture\_as\_pdf2201.08262.pdf](/sites/g/files/omnuum6171/files/crcs/files/2201.08262.pdf) Assessing the effects of a policy based on observational data from a different policy is a common problem across several high-stake decision-making domains, and several off-policy evaluation (OPE) techniques have been proposed. However, these methods... - [ descriptionPublisher's Version](https://arxiv.org/abs/2201.08262) - [ picture\_as\_pdf2201.08262.pdf](/sites/g/files/omnuum6171/files/crcs/files/2201.08262.pdf) Killian, T., Ghassemi & Joshi. [Counterfactually Guided Off-policy Transfer in Clinical Settings](/publications/counterfactually-guided-policy-transfer-clinical-settings). in *Conference for Health, Inference, and Learning (CHIL) 2022*. Killian, T., Ghassemi & Joshi. [Counterfactually Guided Off-policy Transfer in Clinical Settings](/publications/counterfactually-guided-policy-transfer-clinical-settings). in *Conference for Health, Inference, and Learning (CHIL) 2022*. - [ descriptionPublisher's Version](https://arxiv.org/pdf/2006.11654.pdf) - [ picture\_as\_pdf2006.11654.pdf](/sites/g/files/omnuum6171/files/crcs/files/2006.11654.pdf) - [ descriptionPublisher's Version](https://arxiv.org/pdf/2006.11654.pdf) - [ picture\_as\_pdf2006.11654.pdf](/sites/g/files/omnuum6171/files/crcs/files/2006.11654.pdf) ### 2024 Karusala, N., Upadhyay, S., Veeraraghavan, R. & Gajos, K. [Understanding Contestability on the Margins: Implications for the Design of Algorithmic Decision-making in Public Services](/publications/understanding-contestability-margins-implications-design-algorithmic-decision). in *CHI Conference on Human Factors in Computing Systems (CHI ’24)* (2024). Karusala, N., Upadhyay, S., Veeraraghavan, R. & Gajos, K. [Understanding Contestability on the Margins: Implications for the Design of Algorithmic Decision-making in Public Services](/publications/understanding-contestability-margins-implications-design-algorithmic-decision). in *CHI Conference on Human Factors in Computing Systems (CHI ’24)* (2024). - add\_circle\_outline do\_not\_disturb\_on Abstract - [ descriptionPublisher's Version](https://dl.acm.org/doi/full/10.1145/3613904.3641898) - [ picture\_as\_pdfkarusala2024understanding...](/sites/g/files/omnuum6171/files/karusala2024understanding.pdf) Policymakers have established that the ability to contest decisions made by or with algorithms is core to responsible artificial intelligence (AI). However, there has been a disconnect between research on contestability of algorithms, and what the... - [ descriptionPublisher's Version](https://dl.acm.org/doi/full/10.1145/3613904.3641898) - [ picture\_as\_pdfkarusala2024understanding...](/sites/g/files/omnuum6171/files/karusala2024understanding.pdf) ### 2023 Ehrmann, D. E., Joshi, S., Goodfellow, S. D., Mazwi, M. L. & Eytan, D. [Making machine learning matter to clinicians: model actionability in medical decision-making](/publications/making-machine-learning-matter-clinicians-model-actionability-medical-decision). *NPJ Digital Medicine* **6**, 7 (2023). Ehrmann, D. E., Joshi, S., Goodfellow, S. D., Mazwi, M. L. & Eytan, D. [Making machine learning matter to clinicians: model actionability in medical decision-making](/publications/making-machine-learning-matter-clinicians-model-actionability-medical-decision). *NPJ Digital Medicine* **6**, 7 (2023). - add\_circle\_outline do\_not\_disturb\_on Abstract - [ descriptionPublisher's Version](https://www.nature.com/articles/s41746-023-00753-7) - [ picture\_as\_pdfs41746-023-00753-7.pdf](/sites/g/files/omnuum6171/files/crcs/files/s41746-023-00753-7.pdf) Machine learning (ML) has the potential to transform patient care and outcomes. However, there are important differences between measuring the performance of ML models in silico and usefulness at the point of care. One lens to use to evaluate models... - [ descriptionPublisher's Version](https://www.nature.com/articles/s41746-023-00753-7) - [ picture\_as\_pdfs41746-023-00753-7.pdf](/sites/g/files/omnuum6171/files/crcs/files/s41746-023-00753-7.pdf) ### 2022 Pawelczyk, M., Agarwal, C., Joshi, S., Upadhyay, S. & Lakkaraju, H. [Exploring Counterfactual Explanations through the lens of Adversarial Examples: A Theoretical and Empirical Analysis.](/publications/exploring-counterfactual-explanations-through-lens-adversarial-examples) *International Conference on Artificial Intelligence and Statistics (AISTATS)* (2022). Pawelczyk, M., Agarwal, C., Joshi, S., Upadhyay, S. & Lakkaraju, H. [Exploring Counterfactual Explanations through the lens of Adversarial Examples: A Theoretical and Empirical Analysis.](/publications/exploring-counterfactual-explanations-through-lens-adversarial-examples) *International Conference on Artificial Intelligence and Statistics (AISTATS)* (2022). - add\_circle\_outline do\_not\_disturb\_on Abstract - [ descriptionPublisher's Version](https://arxiv.org/abs/2106.09992) - [ picture\_as\_pdf2106.09992.pdf](/sites/g/files/omnuum6171/files/crcs/files/2106.09992.pdf) As machine learning (ML) models become more widely deployed in high-stakes applications, counterfactual explanations have emerged as key tools for providing actionable model explanations in practice. Despite the growing popularity of counterfactual... - [ descriptionPublisher's Version](https://arxiv.org/abs/2106.09992) - [ picture\_as\_pdf2106.09992.pdf](/sites/g/files/omnuum6171/files/crcs/files/2106.09992.pdf) ### 2021 Zhang, H. *et al.* [An Empirical Framework for Domain Generalization in Clinical Settings](/publications/empirical-framework-domain-generalization-clinical-settings). *Conference for Health, Inference, and Learning (CHIL) 2021* (2021). Zhang, H. *et al.* [An Empirical Framework for Domain Generalization in Clinical Settings](/publications/empirical-framework-domain-generalization-clinical-settings). *Conference for Health, Inference, and Learning (CHIL) 2021* (2021). - add\_circle\_outline do\_not\_disturb\_on Abstract - [ descriptionPublisher's Version](https://arxiv.org/abs/2103.11163) - [ picture\_as\_pdf2103.11163.pdf](/sites/g/files/omnuum6171/files/crcs/files/2103.11163.pdf) Clinical machine learning models experience significantly degraded performance in datasets not seen during training, e.g., new hospitals or populations. Recent developments in domain generalization offer a promising solution to this problem by... - [ descriptionPublisher's Version](https://arxiv.org/abs/2103.11163) - [ picture\_as\_pdf2103.11163.pdf](/sites/g/files/omnuum6171/files/crcs/files/2103.11163.pdf) [ More arrow\_circle\_right ](/publications)