Demba Kande
Demba Kande has a total of 19 co-authors. Here are the top co-authors by number of shared publications:
- Abdoulaye Déthié Sarr
University of Alioune Diop Bambey, Senegal
1 shared publication Abiboulaye Sall
Affiliation PATH Malaria Control and Elimination Partnership in Africa, Dakar, Senegal
1 shared publicationAdam Bennett
Affiliations Malaria Elimination Initiative, Institute for Global Health Sciences, University of California San Francisco, San Francisco, California, United States of America, PATH Malaria Control and Elimination Partnership in Africa, Seattle, Washington, United States of America
1 shared publicationAichatou Barry Diouf
Affiliation Ministere de la santé et de l’action sociale, Dakar, Senegal
1 shared publication- Baba Camara
Affiliation Ministere de la santé et de l’action sociale, Dakar, Senegal
1 shared publication - Bryan Greenhouse
Department of Medicine, University of California, San Francisco, USA
1 shared publication
Latest Publications
- Open Access
- 29 April 2025
- Published by Public Library of Science (PLoS) in PLOS Global Public Health
- Vol. 5 (4) , e0004569
Feasibility and effectiveness of tailored interventions for two populations at high-risk of malaria in Senegal: Koranic school children and gold miners
Senegal has made significant progress in reducing the malaria burden over the last decade. However, malaria remains a major cause of morbidity in some regions and key challenges exist among high-risk populations who have high exposure to mosquitos, but low coverage and use of vector control measures and limited access to healthcare. Two identified high-risk populations are goldminers and talibés (Koranic school students). We conducted a controlled pre/post survey to measure the impact of targeted malaria interventions, including expansion of active community case management and distribution of LLINs, on reported LLIN usage and Plasmodium falciparum infection prevalence at mining sites and Koranic schools (daaras) during the high transmission season in Senegal. We randomly assigned four health facility catchment areas in Kaolack (a city with many daaras) and four in Saraya (a district with gold mining sites) to intervention or control groups. Surveys were conducted pre (Oct 2021; n = 1740 talibés and gold miners) and post (Feb 2022; n = 2200) delivery of the intervention package to assess intervention coverage and infection prevalence by rapid diagnostic test and qPCR. We compared infection prevalence and self-reported LLIN usage, by group and arm, between the two time periods using a difference in difference framework with binomial generalized linear mixed models. Among the talibés, the package of interventions was associated with an adjusted 12.6-percentage point relative reduction in RDT-derived malaria prevalence (p < 0.05, adjusted risk difference: -12.6, 95% CI: -2.7, -22.4) and an adjusted 44.0-percentage point increase in reported prior night net use (p < 0.001, aRD: 44.0, 95% CI: 36.3, 51.6) in the intervention group compared to the counterfactual. However, among the gold miners there was no measured association between the package of interventions and these outcomes. While there was high acceptability in both groups, interventions should be tailored to address high mobility amongst gold miners and maximize impact.
- 29 April 2025
- Published by Public Library of Science (PLoS) in PLOS Global Public Health
- Vol. 5 (4) , e0004569
Open AccessFeasibility and effectiveness of tailored interventions for two populations at high-risk of malaria in Senegal: Koranic school children and gold miners
Senegal has made significant progress in reducing the malaria burden over the last decade. However, malaria remains a major cause of morbidity in some regions and key challenges exist among high-risk populations who have high exposure to mosquitos, but low coverage and use of vector control measures and limited access to healthcare. Two identified high-risk populations are goldminers and talibés (Koranic school students). We conducted a controlled pre/post survey to measure the impact of targeted malaria interventions, including expansion of active community case management and distribution of LLINs, on reported LLIN usage and Plasmodium falciparum infection prevalence at mining sites and Koranic schools (daaras) during the high transmission season in Senegal. We randomly assigned four health facility catchment areas in Kaolack (a city with many daaras) and four in Saraya (a district with gold mining sites) to intervention or control groups. Surveys were conducted pre (Oct 2021; n = 1740 talibés and gold miners) and post (Feb 2022; n = 2200) delivery of the intervention package to assess intervention coverage and infection prevalence by rapid diagnostic test and qPCR. We compared infection prevalence and self-reported LLIN usage, by group and arm, between the two time periods using a difference in difference framework with binomial generalized linear mixed models. Among the talibés, the package of interventions was associated with an adjusted 12.6-percentage point relative reduction in RDT-derived malaria prevalence (p < 0.05, adjusted risk difference: -12.6, 95% CI: -2.7, -22.4) and an adjusted 44.0-percentage point increase in reported prior night net use (p < 0.001, aRD: 44.0, 95% CI: 36.3, 51.6) in the intervention group compared to the counterfactual. However, among the gold miners there was no measured association between the package of interventions and these outcomes. While there was high acceptability in both groups, interventions should be tailored to address high mobility amongst gold miners and maximize impact.
- Open Access
- 21 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 71-76
A lexicon-based sentiment analysis approach using a graph structure for modeling relationships between opinion words in French and Wolof corpora
Sentiment analysis, also called opinion mining, is the field of study that analyzes people’s opinions, sentiments, evaluations and emotions from digital text [6]. Many methods have been developed, including the recent development of LLMs. Although these large linguistic models have great potential because they are trained on thousands or millions of gigabytes of text, they perform less well on content in African languages, particularly Wolof. In social networks, Senegalese citizens freely express their opinions using both French and Wolof, often resulting in a large number of spelling errors in their writings. The complexity of Wolof spelling and French grammar and conjugation complicates the analysis of Senegalese citizens’ opinions. In this paper, we address this problem by proposing an approach consisting of a dictionary-based lemmatizer and an algorithm modeling the relationship between French and Wolof opinion words based on a graph structure. Experimental results on the same data [2] show that it successfully predicts sentiment polarity with an F1 score of 85%, compared with 45% for the GPT-3 algorithm and 15% for VADER.
- 21 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 71-76
Open AccessA lexicon-based sentiment analysis approach using a graph structure for modeling relationships between opinion words in French and Wolof corpora
Sentiment analysis, also called opinion mining, is the field of study that analyzes people’s opinions, sentiments, evaluations and emotions from digital text [6]. Many methods have been developed, including the recent development of LLMs. Although these large linguistic models have great potential because they are trained on thousands or millions of gigabytes of text, they perform less well on content in African languages, particularly Wolof. In social networks, Senegalese citizens freely express their opinions using both French and Wolof, often resulting in a large number of spelling errors in their writings. The complexity of Wolof spelling and French grammar and conjugation complicates the analysis of Senegalese citizens’ opinions. In this paper, we address this problem by proposing an approach consisting of a dictionary-based lemmatizer and an algorithm modeling the relationship between French and Wolof opinion words based on a graph structure. Experimental results on the same data [2] show that it successfully predicts sentiment polarity with an F1 score of 85%, compared with 45% for the GPT-3 algorithm and 15% for VADER.