AI-based Diagnostics - Improving Current Diagnostics with AI
- Jun 27, 2024
- 2 min read

AI is transforming healthcare diagnostics by boosting the accuracy and speed of disease detection. These systems can learn from massive medical datasets, spotting patterns and anomalies that human eyes might miss. This is especially crucial for early diagnosis and intervention, vital for conditions like neurological disorders.
At Starlab, we're integrating AI-driven devices and algorithms into everyday clinical practice. Our goal? To provide healthcare professionals with powerful tools for early detection and monitoring of brain diseases. We're committed to making a positive impact with cutting-edge neurotechnology, ensuring early and accurate diagnosis becomes the norm in brain health.
Transformative Diagnostic Studies
In a recent diagnostic study, the AI model SCORE-AI was trained on 30,493 EEGs to distinguish normal from abnormal recordings and classify the latter into epileptiform-focal, epileptiform-generalized, nonepileptiform-focal, and nonepileptiform-diffuse categories. Validated on 9,945 EEGs across three independent datasets, SCORE-AI demonstrated diagnostic accuracy comparable to human experts. This AI-driven approach holds promise for enhancing patient care by boosting efficiency and consistency in specialized epilepsy centers’ processes, highlighting the significant potential for AI in routine clinical EEG interpretation. Read more about SCORE-AI.

Advancements in Parkinson's Disease Diagnostics
At Starlab, we are also making big strides at leveraging the huge potential of AI-based algorithms! We’ve developed a machine learning system to characterize the prodromal phase of Parkinson's Disease using EEG biomarkers (see here). In studies with 118 and 212 participants, we collected EEG data from patients with REM Behavior Disorder (RBD), a potential early sign of Parkinson's Disease (PD) or Dementia with Lewy Bodies (DLB). Our EEG-based machine learning system predicted who would develop PD or DLB with 91% accuracy and an AUC score of 0.82. This reinforces our commitment to revolutionizing neurodegenerative disease diagnosis with cost-effective, non-invasive EEG methods.
Leading the Way in AI-based Healthcare Solutions
We're leveraging our EEG expertise and commitment to AI and machine learning to lead in AI-based healthcare solutions. Our work includes machine and deep learning models for better diagnosis of neurodegenerative diseases like Alzheimer’s and improving clinical trial population stratification. Stay tuned for more updates! 🚀🧠✨
Are you interested in learning how EEG can improve the treatment and monitoring of patients with cognitive complaints at your clinic?
Discover our EEG reporting services: https://www.starlab.es/services/eeg-clinical-reporting
Curious about the impact of EEG as outcome measures for your clinical trials?
Explore our EEG Clinical Support services: https://www.starlab.es/services/clinical-trial-support
Want to learn more?
Get in touch with us at info@starlab.es!




AI-based diagnostics are helping improve the accuracy and efficiency of modern diagnostic processes by supporting faster data analysis, pattern recognition, and decision-making in healthcare and other technical fields. As AI tools continue to evolve, they are increasingly being used to assist professionals in identifying issues earlier and optimizing workflows. These innovation and technology-driven concepts are also discussed in the artificial intelligence (ai) productivity seminar & course for executives in Barcelona, Spain, where professionals explore practical applications of AI in productivity, operations, and strategic decision-making.
He estado revisando últimamente varias herramientas que analizan textos generados por IA, más por curiosidad técnica que por confianza plena. En mi caso suelo fijarme en cómo manejan los matices del lenguaje, porque ahí es donde muchos sistemas fallan. Algunos detectores comparan patrones de redacción demasiado uniformes y estructuras poco naturales. En una de las pruebas que hice encontré una referencia que explicaba el enfoque técnico en esta herramienta hizo que mi texto sonara más natural, y decidí comprobarlo por mi cuenta. Probé con textos mezclados (parte humana, parte generada) y el resultado fue bastante frío: separó segmentos sospechosos sin pedir registro ni demasiados datos. No diría que es infalible, pero al menos el análisis parece más metódico que muchas…