Bridging Gender Disparities in Mental Health: A Bilingual Large Language Model for Multilingual Therapeutic Chatbots

Significant gender disparities and a general lack of support systems compound the state of mental health in developing countries. Women, in particular, face unique challenges due to societal norms and expectations that often restrict their access to mental health services. Additionally, the cultural stigma associated with mental health issues affects both men and women but can be particularly debilitating for women who may be less likely to seek help or disclose their struggles. This study explores the application of Jais-13B, a pre-existing bilingual Large Language Model (LLM) proficient in English and Arabic, tailored to address mental health inquiries. We employed a rich dataset encompassing both languages, essential for fine-tuning and enhancing Jais-13B’s responsiveness and empathetic engagement in mental health contexts. The optimization focused on improving the model’s ability to deliver accurate and contextually aware responses, which is crucial for effective mental health support. Performance evaluation using the cross-entropy loss function demonstrated that Jais-13B achieved an accuracy of 70.58%, indicating its capability to produce relevant and precise answers to mental health questions. These results showcase Jais-13B’s potential to enhance digital mental health interventions significantly, providing accessible care solutions across diverse linguistic landscapes. This study reaffirms the effectiveness of bilingual LLMs in mental health applications and provides insights into their practical implementation, setting a foundational benchmark for future research in multilingual therapeutic chatbots.