Liaquat National Hospital & Medical College

Drug-resistant epilepsy (DRE) still poses one of the greatest therapeutic challenges, afflicting about one-third of all patients with epilepsy in the world. Despite spectacular advances in the fields of anti-seizure medications and neurostimulation techniques, treatment outcomes in DRE have reached plateau levels, signifying an urgent need for better mechanistic understanding and therapeutic strategies. New evidence increasingly elucidates mitochondrial dysfunction as a lens through which to understand seizure generation, pharmacoresistance, and disease progression. Mitochondria are regulators of ATP production, calcium buffering, and redox homeostasis; disruption of any such pathway will result in neuronal hyperexcitability, oxidative injury, and cell death. Moreover, mitochondrial DNA mutations and heteroplasmy threshold can correlate with seizure onset, seizure severity, and Response to treatment, thus being potential biomarkers for risk stratification. This narrative review surveys both preclinical and clinical evidence for mitochondrial dysfunction in epilepsy, examining oxidative stress pathways, mitophagy, and mitochondrial permeability transition pore opening as key mechanisms of neuronal vulnerability. We subsequently analyze various preclinical models of mitochondrial dysfunction, pointing out their respective strengths and weaknesses. Emerging therapeutic strategies, encompassing pharmacological agents, gene therapy, diet, and natural compounds, are then reviewed, which aim to resolve issues surrounding mitochondrial health on a molecular basis. By straddling the mechanistic and clinical narratives, this work foregrounds mitochondrial-centered approaches as promises for both the diagnostic and therapeutic arsenal in the management of DRE.

Artificial intelligence (AI) is reshaping the landscape of attention deficit hyperactivity disorder (ADHD) diagnosis through data-driven and technology-enhanced methodologies. This scoping review, conducted in accordance with PRISMA guidelines, systematically analyzed 54 studies published over the past two decades to assess AI's role in ADHD detection and evaluation. The included studies primarily explored AI applications in brain imaging (MRI), brain activity monitoring (EEG and ECG), behavioral assessments, virtual reality-based testing, and motion-tracking sensors. Among the AI technologies examined, machine learning (ML) and deep learning (DL) algorithms demonstrated promising diagnostic accuracy, with performance rates ranging from 70% to 95%. Convolutional neural networks (CNNs) and support vector machines (SVMs) were particularly effective in image and signal analysis, while natural language processing (NLP) models showed potential in behavioral and cognitive assessments. Despite these advancements, challenges such as algorithmic bias, inconsistent data quality, and the need for extensive, diverse datasets remain barriers to widespread clinical integration. Moreover, while AI models enhance speed and precision in ADHD detection, their applicability in treatment monitoring and personalized intervention remains an area for future research. This review underscores the transformative potential of AI in ADHD diagnosis and advocates for a hybrid approach that integrates AI-driven tools with traditional clinical assessments to enhance diagnostic reliability and patient outcomes.