Development, validation and pilot study of a deep learning based artificial intelligence (AI) model for early detection of leprosy in Indian skin- a mobile application based approach for primary healthcare workers
试验速览
- 阶段
- 不适用
- 状态
- 尚未招募
- 发起方
- 入组人数
- 323
- 试验地点
- 1
- 主要终点
- To know whether the AI deep learning model achieves a minimum performance of more than 70 percent accuracy and area under ROC validate its feasibility for early detection of leprosy in Indian skin.
研究概览
简要总结
Leprosy continues to be a significant public health concern globally, with India being one of the leading contributor. The limited healthcare resources in developing countries often lead to delay in diagnosis, resulting in permanent disabilities and social marginalization. Lack of specialists and limited knowledge regarding leprosy among the primary healthcare workers contribute to a significant delay in diagnosis, revealing the existing gaps between the global leprosy elimination strategies and real world scenarios. In this regard, it is of pivotal importance that we develop alternative approaches to diagnose leprosy cases at an early stage.
Artificial intelligence (AI) is a constantly evolving technology, which has gained significant interest recently. Deep learning is a subset of AI, that utilizes artificial neural networks to autonomously learn from vast amounts of data, enabling models to perform specific tasks without the manual inputs. It has been implemented in diagnosis of different dermatological conditions including leprosy in the recent past with significant success rates. However, there is scarcity of data available regarding utility in Indian skin.
In 2024, World Health Organization had conducted a real-world field study of a skin-NTDs (Neglected Tropical Diseases) mobile application, which uses AI- based models to address 12 NTDs and 24 other common skin conditions .
A similar deep learning model can be trained with feeding clinical images of varied presentations of skin lesions in leprosy and by integrating it with a mobile application, it ensures that the technology is of practical utility. The further validation of the technology by conducting a pilot study among primary health care workers would help assess its scope for real world implementation. Thus, the study aims at developing a deep learning AI model and its validation through a pilot study among primary health care workers.
研究设计
- 研究类型
- Observational
入排标准
- 年龄范围
- 5.00 Year(s) 至 75.00 Year(s)（—）
- 性别
- All
入选标准
- •Patients of all ages and genders with diverse types of leprosy such as the indeterminate, tuberculoid, borderline tuberculoid, borderline boderline, lepromatous leprosy and lepra reactions- type 1 and
- •Patients consenting to participate in the study.
排除标准
- •Patients who were treated for leprosy with anti-leprosy drugs for more than 3 months.
- •Patients with pure neuritic leprosy and diffuse infiltration of LL leprosy.
结局指标
主要结局
To know whether the AI deep learning model achieves a minimum performance of more than 70 percent accuracy and area under ROC validate its feasibility for early detection of leprosy in Indian skin.
时间窗: At 18 months
次要结局
- To know whether the model will be able to acheive more than 80 to 90 percentage accuracy and area under curve which would indicate a robust model with strong clinical applicability and further justifying large scale implementation(1 year 6 months)
研究者
Dr S Chidambara Murthy
Ballari Medical College and Research Centre, Ballari (formerly VIMS)