Daksha Shah
Research Subjects
Daksha Shah has a total of 25 co-authors. Here are the top co-authors by number of shared publications:
- Alex van Belkum
Data Analytics Department, BioMérieux, La Balme les Grottes, France
1 shared publication - Amit Sharma
Microsoft Research India
1 shared publication - Anjali Shetty
Microbiology Section, Department of Laboratory Medicine, P.D. Hinduja National Hospital and Medical Research Centre, Mumbai, India
1 shared publication - Anup Malani
University of Chicago
1 shared publication - Bistra Dilkina
University of Southern California
1 shared publication - Bryan Wilder
University of Southern California
1 shared publication
Latest Publications
- Open Access
Seroprevalence of SARS-CoV-2 in slums and non-slums of Mumbai, India, during June 29-July 19, 2020
Objective: Estimate seroprevalence in representative samples from slum and non-slum communities in Mumbai, India, a mega-city in a low or middle-income country and test if prevalence is different in slums. Design: After geographically-spaced community sampling of households, one individual per household was tested for IgG antibodies to SARS-CoV-2 N-protein in a two-week interval. Setting: Slum and non-slum communities in three wards, one each from the three main zones of Mumbai. Participants: Individuals over age 12 who consent to and have no contraindications to venipuncture were eligible. 6,904 participants (4,202 from slums and 2,702 from non-slums) were tested. Main outcome measures: The primary outcomes were the positive test rate for IgG antibodies to the SARS-CoV-2 N-protein by demographic group (age and gender) and location (slums and non-slums). The secondary outcome is seroprevalence at slum and non-slum levels. Sera was tested via chemiluminescence (CLIA) using Abbott Diagnostics Architect TM N-protein based test. Seroprevalence was calculated using weights to match the population distribution by age and gender and accounting for imperfect sensitivity and specificity of the test. Results: The positive test rate was 54.1% (95% CI: 52.7 to 55.6) and 16.1% (95% CI: 14.9 to 17.4) in slums and non-slums, respectively, a difference of 38 percentage points (P < 0.001). Accounting for imperfect accuracy of tests (e.g., sensitivity, 0.90; specificity 1.00), seroprevalence was as high as 58.4% (95% CI: 56.8 to 59.9) and 17.3% (95% CI: 16 to 18.7) in slums and non-slums, respectively. Conclusions: The high seroprevalence in slums implies a moderate infection fatality rate. The stark difference in seroprevalence across slums and non-slums has implications for the efficacy of social distancing, the level of herd immunity, and equity. It underlines the importance of geographic specificity and urban structure in modeling SARS-CoV-2.
Open AccessSeroprevalence of SARS-CoV-2 in slums and non-slums of Mumbai, India, during June 29-July 19, 2020
Objective: Estimate seroprevalence in representative samples from slum and non-slum communities in Mumbai, India, a mega-city in a low or middle-income country and test if prevalence is different in slums. Design: After geographically-spaced community sampling of households, one individual per household was tested for IgG antibodies to SARS-CoV-2 N-protein in a two-week interval. Setting: Slum and non-slum communities in three wards, one each from the three main zones of Mumbai. Participants: Individuals over age 12 who consent to and have no contraindications to venipuncture were eligible. 6,904 participants (4,202 from slums and 2,702 from non-slums) were tested. Main outcome measures: The primary outcomes were the positive test rate for IgG antibodies to the SARS-CoV-2 N-protein by demographic group (age and gender) and location (slums and non-slums). The secondary outcome is seroprevalence at slum and non-slum levels. Sera was tested via chemiluminescence (CLIA) using Abbott Diagnostics Architect TM N-protein based test. Seroprevalence was calculated using weights to match the population distribution by age and gender and accounting for imperfect sensitivity and specificity of the test. Results: The positive test rate was 54.1% (95% CI: 52.7 to 55.6) and 16.1% (95% CI: 14.9 to 17.4) in slums and non-slums, respectively, a difference of 38 percentage points (P < 0.001). Accounting for imperfect accuracy of tests (e.g., sensitivity, 0.90; specificity 1.00), seroprevalence was as high as 58.4% (95% CI: 56.8 to 59.9) and 17.3% (95% CI: 16 to 18.7) in slums and non-slums, respectively. Conclusions: The high seroprevalence in slums implies a moderate infection fatality rate. The stark difference in seroprevalence across slums and non-slums has implications for the efficacy of social distancing, the level of herd immunity, and equity. It underlines the importance of geographic specificity and urban structure in modeling SARS-CoV-2.
- Open Access
Learning to Prescribe Interventions for Tuberculosis Patients Using Digital Adherence Data
Digital Adherence Technologies (DATs) are an increasingly popular method for verifying patient adherence to many medications. We analyze data from one city served by 99DOTS, a phone-call-based DAT deployed for Tuberculosis (TB) treatment in India where nearly 3 million people are afflicted with the disease each year. The data contains nearly 17,000 patients and 2.1M dose records. We lay the groundwork for learning from this real-world data, including a method for avoiding the effects of unobserved interventions in training data used for machine learning. We then construct a deep learning model, demonstrate its interpretability, and show how it can be adapted and trained in different clinical scenarios to better target and improve patient care. In the real-time risk prediction setting our model could be used to proactively intervene with 21% more patients and before 76% more missed doses than current heuristic baselines. For outcome prediction, our model performs 40% better than baseline methods, allowing cities to target more resources to clinics with a heavier burden of patients at risk of failure. Finally, we present a case study demonstrating how our model can be trained in an end-to-end decision focused learning setting to achieve 15% better solution quality in an example decision problem faced by health workers.
Open AccessLearning to Prescribe Interventions for Tuberculosis Patients Using Digital Adherence Data
Digital Adherence Technologies (DATs) are an increasingly popular method for verifying patient adherence to many medications. We analyze data from one city served by 99DOTS, a phone-call-based DAT deployed for Tuberculosis (TB) treatment in India where nearly 3 million people are afflicted with the disease each year. The data contains nearly 17,000 patients and 2.1M dose records. We lay the groundwork for learning from this real-world data, including a method for avoiding the effects of unobserved interventions in training data used for machine learning. We then construct a deep learning model, demonstrate its interpretability, and show how it can be adapted and trained in different clinical scenarios to better target and improve patient care. In the real-time risk prediction setting our model could be used to proactively intervene with 21% more patients and before 76% more missed doses than current heuristic baselines. For outcome prediction, our model performs 40% better than baseline methods, allowing cities to target more resources to clinics with a heavier burden of patients at risk of failure. Finally, we present a case study demonstrating how our model can be trained in an end-to-end decision focused learning setting to achieve 15% better solution quality in an example decision problem faced by health workers.
- 1 May 2018
- Published by Elsevier in Tuberculosis
- Vol. 110, 86-90
Evaluation of pyrosequencing for extensive drug resistance-defining anti-tuberculosis drugs for use in public healthcare
- 1 May 2018
- Published by Elsevier in Tuberculosis
- Vol. 110, 86-90
Evaluation of pyrosequencing for extensive drug resistance-defining anti-tuberculosis drugs for use in public healthcare