Journal Article > ResearchFull Text
PLoS Negl Trop Dis. 21 June 2024; Volume 18 (Issue 6); e0011978.; DOI:10.1371/journal.pntd.0011978
Musumeci S, Kruse A, Chappuis F, Ostergaard Jensen T, Alcoba G
PLoS Negl Trop Dis. 21 June 2024; Volume 18 (Issue 6); e0011978.; DOI:10.1371/journal.pntd.0011978
BACKGROUND
Febrile illnesses that persist despite initial treatment are common clinical challenges in (sub)tropical low-resource settings. Our aim is to review infectious etiologies of “prolonged fevers” (persistent febrile illnesses, PFI) and to quantify relative contributions of selected neglected target diseases with limited diagnostic options, often overlooked, causing inadequate antibiotic prescriptions, or requiring prolonged and potentially toxic treatments.
METHODS
We performed a systematic review of articles addressing the infectious etiologies of PFI in adults and children in sub-/tropical low- and middle-income countries (LMICs) using the PRISMA guidelines. A list of target diseases, including neglected parasites and zoonotic bacteria (e.g., Leishmania and Brucella), were identified by infectious diseases and tropical medicine specialists and prioritized in the search. Malaria and tuberculosis (TB) were not included as target diseases due to well-established epidemiology and diagnostic options. Four co-investigators independently extracted data from the identified articles while assessing for risk of bias.
RESULTS
196 articles from 52 countries were included, 117 from Africa (33 countries), 71 from Asia (16 countries), and 8 from Central and -South America (3 countries). Target diseases were reported as the cause of PFI in almost half of the articles, most frequently rickettsioses (including scrub typhus), relapsing fever borreliosis (RF-borreliosis), brucellosis, enteric fever, leptospirosis, Q fever and leishmaniasis. Among those, RF-borreliosis was by far the most frequently reported disease in Africa, particularly in Eastern Africa. Rickettsioses (including scrub typhus) were often described in both Africa and Asia. Leishmaniasis, toxoplasmosis and amoebiasis were the most frequent parasitic etiologies. Non-target diseases and non-tropical organisms (Streptococcus pneumoniae, Escherichia coli, and non-typhoidal Salmonella spp) were documented in a fifth of articles.
CONCLUSIONS
Clinicians faced with PFI in sub-/tropical LMICs should consider a wide differential diagnosis including enteric fever and zoonotic bacterial diseases (e.g., rickettsiosis, RF-borreliosis and brucellosis), or parasite infections (e.g., leishmaniasis) depending on geography and syndromes. In the absence of adequate diagnostic capacity, a trial of antibiotics targeting relevant intra-cellular bacteria, such as doxycycline or azithromycin, may be considered.
Febrile illnesses that persist despite initial treatment are common clinical challenges in (sub)tropical low-resource settings. Our aim is to review infectious etiologies of “prolonged fevers” (persistent febrile illnesses, PFI) and to quantify relative contributions of selected neglected target diseases with limited diagnostic options, often overlooked, causing inadequate antibiotic prescriptions, or requiring prolonged and potentially toxic treatments.
METHODS
We performed a systematic review of articles addressing the infectious etiologies of PFI in adults and children in sub-/tropical low- and middle-income countries (LMICs) using the PRISMA guidelines. A list of target diseases, including neglected parasites and zoonotic bacteria (e.g., Leishmania and Brucella), were identified by infectious diseases and tropical medicine specialists and prioritized in the search. Malaria and tuberculosis (TB) were not included as target diseases due to well-established epidemiology and diagnostic options. Four co-investigators independently extracted data from the identified articles while assessing for risk of bias.
RESULTS
196 articles from 52 countries were included, 117 from Africa (33 countries), 71 from Asia (16 countries), and 8 from Central and -South America (3 countries). Target diseases were reported as the cause of PFI in almost half of the articles, most frequently rickettsioses (including scrub typhus), relapsing fever borreliosis (RF-borreliosis), brucellosis, enteric fever, leptospirosis, Q fever and leishmaniasis. Among those, RF-borreliosis was by far the most frequently reported disease in Africa, particularly in Eastern Africa. Rickettsioses (including scrub typhus) were often described in both Africa and Asia. Leishmaniasis, toxoplasmosis and amoebiasis were the most frequent parasitic etiologies. Non-target diseases and non-tropical organisms (Streptococcus pneumoniae, Escherichia coli, and non-typhoidal Salmonella spp) were documented in a fifth of articles.
CONCLUSIONS
Clinicians faced with PFI in sub-/tropical LMICs should consider a wide differential diagnosis including enteric fever and zoonotic bacterial diseases (e.g., rickettsiosis, RF-borreliosis and brucellosis), or parasite infections (e.g., leishmaniasis) depending on geography and syndromes. In the absence of adequate diagnostic capacity, a trial of antibiotics targeting relevant intra-cellular bacteria, such as doxycycline or azithromycin, may be considered.
Conference Material > Slide Presentation
Solomos A, Musa AM, Mbui J, Mohammed R, Olobo J, et al.
MSF Scientific Day International 2023. 7 June 2023; DOI:10.57740/81jj-tz57
Conference Material > Abstract
Solomos A, Musa AM, Mbui J, Mohammed R, Olobo J, et al.
MSF Scientific Day International 2023. 7 June 2023; DOI:10.57740/y4z2-pq54
INTRODUCTION
The parasitic disease visceral leishmaniasis (VL) is most commonly caused by Leishmania donovani in eastern Africa, currently the region with highest burden worldwide. Current treatment for VL comprises the combination of sodium stibogluconate and paromomycin, SSG/PM; this is toxic, painful, and requires hospitalisation and daily injections. Treatments are urgently needed that are safe, effective, and appropriate for use in remote areas.
METHODS
We conducted a phase III open-label randomised non-inferiority trial in Ethiopia, Kenya, Sudan, and Uganda. The trial used a parallel-arm design with two arms, comparing the combination regimen of 20 mg/kg/day paromomycin and allometric miltefosine (MF) for 14 days with the current 17-day standard of care, 20 mg/kg/day SSG and 15 mg/kg/day PM. We enrolled adults and children aged 4-50 years with primary VL, without HIV or severe concomitant disease co-infection. The primary endpoint was definitive cure at 6 months’ follow-up.
ETHICS
This study was approved by the MSF Ethics Review Board and by ethics committees at the Institute of Endemic Diseases, Khartoum, Sudan; Kenya Medical Research Institute, Nairobi, Kenya; Makerere University, Uganda; and the University of Gondar, Ethiopia. Clinicaltrials.gov registry number, NCT03129646.
RESULTS
439 predominantly male (80%) patients aged 4 to 50 years were recruited over a period of 29 months. A similar proportion of patients in the PM/MF and the SSG/PM arms achieved definitive cure at 6-month follow-up in primary efficacy analysis using modified intention-to-treat; mITT; 91.2% cure for PM/MF (97.5% confidence interval, CI, 85-98.6) and 91.8% for SSG/PM (97.5% CI, 85.6-99.2). Non-inferiority was not demonstrated in the mITT population, with the upper limit of the 97.5% CI, 7.4%, slightly exceeding the non-inferiority margin of 7%. However, the per protocol analysis did show non-inferiority, with 92% (97.5% CI, 85-98.5) cure in the PM/MF arm, as compared to 91.7% (97.5% CI, 84.7-98.2) in the SSG/PM arm. Most adverse drug reactions (ADR’s) were mild to moderate. The most common expected ADR’s were MF-related vomiting, and PM-related injection site pain and hypoacusis. ADR’s suggesting SSG-related cardiac toxicity were reported in 6.5% (11/170) of patients in the SSG/PM arm. Eighteen serious adverse events were reported in 13 patients, four of which were considered related to study drugs. Fatality rate in the trial was 0.9% (4/439), with one death judged due to SSG-related cardiotoxicity.
CONCLUSION
The results of this study demonstrate that the 14-day PM/MF regimen achieved a clinically meaningful rate of cure with very similar efficacy to the standard of care, SSG/PM. It was generally well tolerated, with ADR’s as expected, based on the known safety profiles of study drugs. The PM/MF regimen has one fewer painful injection per day, a 3-day shorter treatment duration, and with no risk of SSG-associated life threating cardiotoxicity, as compared to SSG/PM. This regimen may therefore provide a more patient-friendly alternative for adults and children with VL in eastern Africa.
CONFLICTS OF INTEREST
None declared
The parasitic disease visceral leishmaniasis (VL) is most commonly caused by Leishmania donovani in eastern Africa, currently the region with highest burden worldwide. Current treatment for VL comprises the combination of sodium stibogluconate and paromomycin, SSG/PM; this is toxic, painful, and requires hospitalisation and daily injections. Treatments are urgently needed that are safe, effective, and appropriate for use in remote areas.
METHODS
We conducted a phase III open-label randomised non-inferiority trial in Ethiopia, Kenya, Sudan, and Uganda. The trial used a parallel-arm design with two arms, comparing the combination regimen of 20 mg/kg/day paromomycin and allometric miltefosine (MF) for 14 days with the current 17-day standard of care, 20 mg/kg/day SSG and 15 mg/kg/day PM. We enrolled adults and children aged 4-50 years with primary VL, without HIV or severe concomitant disease co-infection. The primary endpoint was definitive cure at 6 months’ follow-up.
ETHICS
This study was approved by the MSF Ethics Review Board and by ethics committees at the Institute of Endemic Diseases, Khartoum, Sudan; Kenya Medical Research Institute, Nairobi, Kenya; Makerere University, Uganda; and the University of Gondar, Ethiopia. Clinicaltrials.gov registry number, NCT03129646.
RESULTS
439 predominantly male (80%) patients aged 4 to 50 years were recruited over a period of 29 months. A similar proportion of patients in the PM/MF and the SSG/PM arms achieved definitive cure at 6-month follow-up in primary efficacy analysis using modified intention-to-treat; mITT; 91.2% cure for PM/MF (97.5% confidence interval, CI, 85-98.6) and 91.8% for SSG/PM (97.5% CI, 85.6-99.2). Non-inferiority was not demonstrated in the mITT population, with the upper limit of the 97.5% CI, 7.4%, slightly exceeding the non-inferiority margin of 7%. However, the per protocol analysis did show non-inferiority, with 92% (97.5% CI, 85-98.5) cure in the PM/MF arm, as compared to 91.7% (97.5% CI, 84.7-98.2) in the SSG/PM arm. Most adverse drug reactions (ADR’s) were mild to moderate. The most common expected ADR’s were MF-related vomiting, and PM-related injection site pain and hypoacusis. ADR’s suggesting SSG-related cardiac toxicity were reported in 6.5% (11/170) of patients in the SSG/PM arm. Eighteen serious adverse events were reported in 13 patients, four of which were considered related to study drugs. Fatality rate in the trial was 0.9% (4/439), with one death judged due to SSG-related cardiotoxicity.
CONCLUSION
The results of this study demonstrate that the 14-day PM/MF regimen achieved a clinically meaningful rate of cure with very similar efficacy to the standard of care, SSG/PM. It was generally well tolerated, with ADR’s as expected, based on the known safety profiles of study drugs. The PM/MF regimen has one fewer painful injection per day, a 3-day shorter treatment duration, and with no risk of SSG-associated life threating cardiotoxicity, as compared to SSG/PM. This regimen may therefore provide a more patient-friendly alternative for adults and children with VL in eastern Africa.
CONFLICTS OF INTEREST
None declared
Journal Article > ResearchFull Text
Lancet Global Health. 1 February 2023; Volume 11 (Issue 2); e296-e300.; DOI:10.1016/S2214-109X(22)00479-X
Abouyannis M, Esmail H, Hamaluba M, Ngama M, Mwangudzah H, et al.
Lancet Global Health. 1 February 2023; Volume 11 (Issue 2); e296-e300.; DOI:10.1016/S2214-109X(22)00479-X
Snakebite clinical trials have often used heterogeneous outcome measures and there is an urgent need for standardisation. A globally representative group of key stakeholders came together to reach consensus on a globally relevant set of core outcome measurements. Outcome domains and outcome measurement instruments were identified through searching the literature and a systematic review of snakebite clinical trials. Outcome domains were shortlisted by use of a questionnaire and consensus was reached among stakeholders and the patient group through facilitated discussions and voting. Five universal core outcome measures should be included in all future snakebite clinical trials-mortality, WHO disability assessment scale, patient-specific functional scale, acute allergic reaction by Brown criteria, and serum sickness by formal criteria. Additional syndrome-specific core outcome measures should be used depending on the biting species. This core outcome measurement set provides global standardisation, supports the priorities of patients and clinicians, enables meta-analysis, and is appropriate for use in low-income and middle-income settings.
Journal Article > ResearchFull Text
Med Trop Sante Int
MTSI
MTSI bulletin
MTSI magazine
MTSI-Revue
Médecine tropicale et santé internationale. Bulletin
Médecine tropicale et santé internationale. Magazine
MTSI, la revue de la Société francophone de médecine tropicale et santé internationale. 1 February 2023; Volume 3 (Issue 3); DOI:10.48327/mtsi.v3i3.2023.421
Abouyannis M, Esmail H, Hamaluba M, Ngama M, Mwangudzah H, et al.
Med Trop Sante Int
MTSI
MTSI bulletin
MTSI magazine
MTSI-Revue
Médecine tropicale et santé internationale. Bulletin
Médecine tropicale et santé internationale. Magazine
MTSI, la revue de la Société francophone de médecine tropicale et santé internationale. 1 February 2023; Volume 3 (Issue 3); DOI:10.48327/mtsi.v3i3.2023.421
English
Français
CONTEXTE
Les essais cliniques sur les morsures de serpent ont souvent utilisé des critères de décision hétérogènes qui demandent à être standardisés.
MÉTHODE
Un groupe d'acteurs clés mondialement représentatifs s'est réuni pour parvenir à un consensus sur un jeu universel de critères de décision de base. Les domaines d'intérêt et les instruments d'évaluation des critères de décision ont été identifiés à partir d'une recherche documentaire et d'un examen systématique des essais cliniques concernant les envenimations par morsure de serpent. Les domaines d'intérêt ont été présélectionnés à l'aide d'un questionnaire et un consensus a été obtenu entre le groupe d'acteurs et un groupe représentatif de patients à la suite de discussions orientées et d'un vote.
RÉSULTATS
Cinq critères de décision de base universels devraient être inclus dans tous les futurs essais cliniques sur les morsures de serpent : la mortalité, l'échelle d'évaluation du handicap de l'OMS, l'échelle fonctionnelle propre à chaque patient, la réaction allergique immédiate selon les critères de Brown et la maladie sérique en fonction de critères formels. D'autres critères de décision spécifiques aux différents syndromes observés lors des envenimations par morsure de serpent doivent être utilisés en fonction de l'espèce responsable de la morsure.
CONCLUSION
Ce jeu universel de critères de décision de base permet une standardisation mondiale, répond aux priorités des patients et des cliniciens, favorise des méta-analyses et est compatible avec une utilisation dans les pays à revenu faible ou intermédiaire.
Les essais cliniques sur les morsures de serpent ont souvent utilisé des critères de décision hétérogènes qui demandent à être standardisés.
MÉTHODE
Un groupe d'acteurs clés mondialement représentatifs s'est réuni pour parvenir à un consensus sur un jeu universel de critères de décision de base. Les domaines d'intérêt et les instruments d'évaluation des critères de décision ont été identifiés à partir d'une recherche documentaire et d'un examen systématique des essais cliniques concernant les envenimations par morsure de serpent. Les domaines d'intérêt ont été présélectionnés à l'aide d'un questionnaire et un consensus a été obtenu entre le groupe d'acteurs et un groupe représentatif de patients à la suite de discussions orientées et d'un vote.
RÉSULTATS
Cinq critères de décision de base universels devraient être inclus dans tous les futurs essais cliniques sur les morsures de serpent : la mortalité, l'échelle d'évaluation du handicap de l'OMS, l'échelle fonctionnelle propre à chaque patient, la réaction allergique immédiate selon les critères de Brown et la maladie sérique en fonction de critères formels. D'autres critères de décision spécifiques aux différents syndromes observés lors des envenimations par morsure de serpent doivent être utilisés en fonction de l'espèce responsable de la morsure.
CONCLUSION
Ce jeu universel de critères de décision de base permet une standardisation mondiale, répond aux priorités des patients et des cliniciens, favorise des méta-analyses et est compatible avec une utilisation dans les pays à revenu faible ou intermédiaire.
Journal Article > CommentaryFull Text
Toxicon: X. 21 December 2022; Volume 17; 100146.; DOI:10.1016/j.toxcx.2022.100146
Potet J, Singh SN, Ritmeijer KKD, Sisay K, Alcoba G, et al.
Toxicon: X. 21 December 2022; Volume 17; 100146.; DOI:10.1016/j.toxcx.2022.100146
The medical humanitarian organization Médecins Sans Frontières (MSF) provides medical care in more than 70 countries and admits more than 7000 cases of snakebite in its facilities each year.
We describe our activities against snakebite in three African countries: Central African Republic, South Sudan and Ethiopia, in which different models of care have been developed. A standard protocol using two different antivenoms depending on the patient's syndrome has been introduced, and a simple blood coagulation test is performed to detect venom-induced coagulopathy. Other services, including surgery for necrotizing wounds, are offered in the facilities where MSF admits a large number of snakebite patients. All services, including provision of antivenom, are offered free-of-charge in MSF-supported facilities. Community-based activities focusing on preventive measures and prompt transport to hospital have been developed in a few MSF projects.
The provision of quality care and treatment, including effective antivenoms, without out-of-pocket payments by the patients, probably explains why MSF has admitted an increasing number of snakebite victims over the last years. This model requires significant resources and monitoring, including regular training of healthcare workers on treatment protocols and a considerable budget for antivenom procurement.
We describe our activities against snakebite in three African countries: Central African Republic, South Sudan and Ethiopia, in which different models of care have been developed. A standard protocol using two different antivenoms depending on the patient's syndrome has been introduced, and a simple blood coagulation test is performed to detect venom-induced coagulopathy. Other services, including surgery for necrotizing wounds, are offered in the facilities where MSF admits a large number of snakebite patients. All services, including provision of antivenom, are offered free-of-charge in MSF-supported facilities. Community-based activities focusing on preventive measures and prompt transport to hospital have been developed in a few MSF projects.
The provision of quality care and treatment, including effective antivenoms, without out-of-pocket payments by the patients, probably explains why MSF has admitted an increasing number of snakebite victims over the last years. This model requires significant resources and monitoring, including regular training of healthcare workers on treatment protocols and a considerable budget for antivenom procurement.
Journal Article > ResearchFull Text
Clin Infect Dis. 27 September 2022; Online ahead of print; ciac643.; DOI:10.1093/cid/ciac643
Musa AM, Mbui J, Mohammed R, Olobo J, Ritmeijer KKD, et al.
Clin Infect Dis. 27 September 2022; Online ahead of print; ciac643.; DOI:10.1093/cid/ciac643
BACKGROUND
This study aimed to determine whether paromomycin plus miltefosine (PM/MF) is noninferior to sodium stibogluconate plus paromomycin (SSG/PM) for treatment of primary visceral leishmaniasis in eastern Africa.
METHODS
An open-label, phase 3, randomized, controlled trial was conducted in adult and pediatric patients at 7 sites in eastern Africa. Patients were randomly assigned to either 20 mg/kg paromomycin plus allometric dose of miltefosine (14 days), or 20 mg/kg sodium stibogluconate plus 15 mg/kg paromomycin (17 days). The primary endpoint was definitive cure after 6 months.
RESULTS
Of 439 randomized patients, 424 completed the trial. Definitive cure at 6 months was 91.2% (155 of 170) and 91.8% (156 of 170) in the PM/MF and SSG/PM arms in primary efficacy modified intention-to-treat analysis (difference, 0.6%; 97.5% confidence interval [CI], -6.2 to 7.4), narrowly missing the noninferiority margin of 7%. In the per-protocol analysis, efficacy was 92% (149 of 162) and 91.7% (155 of 169) in the PM/MF and SSG/PM arms (difference, -0.3%; 97.5% CI, –7.0 to 6.5), demonstrating noninferiority. Treatments were well tolerated. Four of 18 serious adverse events were study drug–related, and 1 death was SSG-related. Allometric dosing ensured similar MF exposure in children (< 12 years) and adults.
CONCLUSIONS
PM/MF and SSG/PM efficacies were similar, and adverse drug reactions were as expected given the drugs safety profiles. With 1 less injection each day, reduced treatment duration, and no risk of SSG-associated life-threatening cardiotoxicity, PM/MF is a more patient-friendly alternative for children and adults with primary visceral leishmaniasis in eastern Africa.
CLINICAL TRIALS REGISTRATION
NCT03129646.
This study aimed to determine whether paromomycin plus miltefosine (PM/MF) is noninferior to sodium stibogluconate plus paromomycin (SSG/PM) for treatment of primary visceral leishmaniasis in eastern Africa.
METHODS
An open-label, phase 3, randomized, controlled trial was conducted in adult and pediatric patients at 7 sites in eastern Africa. Patients were randomly assigned to either 20 mg/kg paromomycin plus allometric dose of miltefosine (14 days), or 20 mg/kg sodium stibogluconate plus 15 mg/kg paromomycin (17 days). The primary endpoint was definitive cure after 6 months.
RESULTS
Of 439 randomized patients, 424 completed the trial. Definitive cure at 6 months was 91.2% (155 of 170) and 91.8% (156 of 170) in the PM/MF and SSG/PM arms in primary efficacy modified intention-to-treat analysis (difference, 0.6%; 97.5% confidence interval [CI], -6.2 to 7.4), narrowly missing the noninferiority margin of 7%. In the per-protocol analysis, efficacy was 92% (149 of 162) and 91.7% (155 of 169) in the PM/MF and SSG/PM arms (difference, -0.3%; 97.5% CI, –7.0 to 6.5), demonstrating noninferiority. Treatments were well tolerated. Four of 18 serious adverse events were study drug–related, and 1 death was SSG-related. Allometric dosing ensured similar MF exposure in children (< 12 years) and adults.
CONCLUSIONS
PM/MF and SSG/PM efficacies were similar, and adverse drug reactions were as expected given the drugs safety profiles. With 1 less injection each day, reduced treatment duration, and no risk of SSG-associated life-threatening cardiotoxicity, PM/MF is a more patient-friendly alternative for children and adults with primary visceral leishmaniasis in eastern Africa.
CLINICAL TRIALS REGISTRATION
NCT03129646.
Journal Article > ResearchFull Text
PLoS Negl Trop Dis. 15 August 2022; Volume 16 (Issue 8); e0010647.; DOI:10.1371/journal.pntd.0010647
Bolon I, Picek L, Durso AM, Alcoba G, Chappuis F, et al.
PLoS Negl Trop Dis. 15 August 2022; Volume 16 (Issue 8); e0010647.; DOI:10.1371/journal.pntd.0010647
BACKGROUND
Snakebite envenoming is a neglected tropical disease that kills an estimated 81,000 to 138,000 people and disables another 400,000 globally every year. The World Health Organization aims to halve this burden by 2030. To achieve this ambitious goal, we need to close the data gap in snake ecology and snakebite epidemiology and give healthcare providers up-to-date knowledge and access to better diagnostic tools. An essential first step is to improve the capacity to identify biting snakes taxonomically. The existence of AI-based identification tools for other animals offers an innovative opportunity to apply machine learning to snake identification and snakebite envenoming, a life-threatening situation.
METHODOLOGY
We developed an AI model based on Vision Transformer, a recent neural network architecture, and a comprehensive snake photo dataset of 386,006 training photos covering 198 venomous and 574 non-venomous snake species from 188 countries. We gathered photos from online biodiversity platforms (iNaturalist and HerpMapper) and a photo-sharing site (Flickr).
PRINCIPAL FINDINGS
The model macro-averaged F1 score, which reflects the species-wise performance as averaging performance for each species, is 92.2%. The accuracy on a species and genus level is 96.0% and 99.0%, respectively. The average accuracy per country is 94.2%. The model accurately classifies selected venomous and non-venomous lookalike species from Southeast Asia and sub-Saharan Africa.
CONCLUSIONS
To our knowledge, this model’s taxonomic and geographic coverage and performance are unprecedented. This model could provide high-speed and low-cost snake identification to support snakebite victims and healthcare providers in low-resource settings, as well as zoologists, conservationists, and nature lovers from across the world.
Snakebite envenoming is a neglected tropical disease that kills an estimated 81,000 to 138,000 people and disables another 400,000 globally every year. The World Health Organization aims to halve this burden by 2030. To achieve this ambitious goal, we need to close the data gap in snake ecology and snakebite epidemiology and give healthcare providers up-to-date knowledge and access to better diagnostic tools. An essential first step is to improve the capacity to identify biting snakes taxonomically. The existence of AI-based identification tools for other animals offers an innovative opportunity to apply machine learning to snake identification and snakebite envenoming, a life-threatening situation.
METHODOLOGY
We developed an AI model based on Vision Transformer, a recent neural network architecture, and a comprehensive snake photo dataset of 386,006 training photos covering 198 venomous and 574 non-venomous snake species from 188 countries. We gathered photos from online biodiversity platforms (iNaturalist and HerpMapper) and a photo-sharing site (Flickr).
PRINCIPAL FINDINGS
The model macro-averaged F1 score, which reflects the species-wise performance as averaging performance for each species, is 92.2%. The accuracy on a species and genus level is 96.0% and 99.0%, respectively. The average accuracy per country is 94.2%. The model accurately classifies selected venomous and non-venomous lookalike species from Southeast Asia and sub-Saharan Africa.
CONCLUSIONS
To our knowledge, this model’s taxonomic and geographic coverage and performance are unprecedented. This model could provide high-speed and low-cost snake identification to support snakebite victims and healthcare providers in low-resource settings, as well as zoologists, conservationists, and nature lovers from across the world.
Journal Article > ResearchFull Text
Am J Trop Med Hyg. 22 February 2018; Volume 98 (Issue 4); 1091–1101.; DOI:10.4269/ajtmh.17-0872
Sunyoto T, Adam GK, Atia AM, Hamid Y, Babiker RA, et al.
Am J Trop Med Hyg. 22 February 2018; Volume 98 (Issue 4); 1091–1101.; DOI:10.4269/ajtmh.17-0872
Early diagnosis and treatment is the principal strategy to control visceral leishmaniasis (VL), or kala-azar in East Africa. As VL strikes remote rural, sparsely populated areas, kala-azar care might not be accessed optimally or timely. We conducted a qualitative study to explore access barriers in a longstanding kala-azar endemic area in southern Gadarif, Sudan. Former kala-azar patients or caretakers, community leaders, and health-care providers were purposively sampled and thematic data analysis was used. Our study participants revealed the multitude of difficulties faced when seeking care. The disease is well known in the area, yet misconceptions about causes and transmission persist. The care-seeking itineraries were not always straightforward: "shopping around" for treatments are common, partly linked to difficulties in diagnosing kala-azar. Kala-azar is perceived to be "hiding," requiring multiple tests and other diseases must be treated first. Negative perceptions on quality of care in the public hospitals prevail, with the unavailability of drugs or staff as the main concern. Delay to seek care remains predominantly linked to economic constraint: albeit treatment is for free, patients have to pay out of pocket for everything else, pushing families further into poverty. Despite increased efforts to tackle the disease over the years, access to quality kala-azar care in this rural Sudanese context remains problematic. The barriers explored in this study are a compelling reminder of the need to boost efforts to address these barriers.
Journal Article > ResearchFull Text
Lancet. 12 July 2018; Volume 392 (Issue 10148); DOI:10.1016/S0140-6736(18)31224-8
Longbottom J, Shearer FM, Devine M, Alcoba G, Chappuis F, et al.
Lancet. 12 July 2018; Volume 392 (Issue 10148); DOI:10.1016/S0140-6736(18)31224-8
Snakebite envenoming is a frequently overlooked cause of mortality and morbidity. Data for snake ecology and existing snakebite interventions are scarce, limiting accurate burden estimation initiatives. Low global awareness stunts new interventions, adequate health resources, and available health care. Therefore, we aimed to synthesise currently available data to identify the most vulnerable populations at risk of snakebite, and where additional data to manage this global problem are needed.