When Innovation Meets Healing
The Biomedical Breakthroughs of ABSET 2023
A Global Convergence for Health
Imagine a future where a stroke patient can regain dexterity through a smart glove and augmented reality, where a simple video can detect your heart rate, or where the complex boundaries of vertebrae are mapped automatically to diagnose spinal conditions. This is not science fiction; it is the tangible present of biomedical engineering, as showcased at the Advances in Biomedical Sciences, Engineering and Technology (ABSET) 2023 international conference.
Global Platform
Connecting scholars and enterprises worldwide
Interdisciplinary
Fusion of computer science, biology, and robotics
Personalized Care
Tailored healthcare solutions for individuals
The Key Concepts: A Symphony of Disciplines
The core of modern biomedical innovation lies in its interdisciplinary nature. The ABSET 2023 conference highlighted several key conceptual pillars that are driving the field forward.
Intelligent Rehabilitation
Moving beyond traditional, repetitive physiotherapy, new systems are leveraging sensor technology and augmented reality (AR) to create adaptive and engaging therapy experiences. These systems can monitor a patient's movements in real time, adjust difficulty, and provide motivating feedback, turning rehabilitation from a chore into an interactive game 1 .
Computational Vision in Healthcare
The ability of computers to "see" and interpret medical images or human features is revolutionizing diagnostics and monitoring. Research at the conference explored everything from automated segmentation of anatomical structures in CT and MRI scans to the analysis of facial expressions for emotion recognition, even when a surgical mask is present 1 .
Data-Driven Personalization
The era of one-size-fits-all medicine is ending. By employing machine learning algorithms on diverse datasets—from skin color videos for heart rate estimation to mechanical gait data—researchers can now develop models that predict outcomes and tailor therapies to the individual's unique physiological makeup 1 .
Non-Invasive Monitoring
A major focus of current research is on developing technologies that gather crucial health data without needles, probes, or complex machinery. Techniques like remote photoplethysmography (rPPG), which can measure heart rate remotely using a standard camera, promise to make continuous health monitoring seamless and integrated into daily life 1 .
A Deep Dive into Rehabotics: The Future of Stroke Recovery
One of the most compelling demonstrations of these converging concepts was the presentation of Rehabotics, an innovative comprehensive platform designed for the rehabilitation of patients with post-stroke spasticity in the upper limbs 1 .
The Experimental Methodology
System Integration
Researchers developed an integrated platform comprising three core components: a sensor-equipped soft glove to capture hand movement and force data, a robotic exoskeleton hand that uses servomotors to assist patients in opening their hand, and an augmented reality (AR) platform with a variety of serious games 1 .
Pilot Study Execution
A pilot study was conducted with a sample of 14 stroke patients to evaluate the system's feasibility and effectiveness 1 .
Data Collection and Therapy Personalization
The soft glove and a web camera collected real-time data on the patient's performance. This data, combined with standard clinical measurements, was used to refine and personalize rehabilitation plans dynamically, targeting the specific motor skills of each individual 1 .
Therapy Sessions
Patients engaged in therapy sessions using the AR serious games. The robotic exoskeleton provided physical assistance where needed, particularly in overcoming the challenge of hand opening, a common impairment after a stroke 1 .
Results and Analysis
The pilot study of the Rehabotics system yielded promising results, underscoring its scientific and clinical importance 1 . The novel system was reported to enhance patient engagement and outcomes in post-stroke spasticity rehabilitation.
Key Finding
By providing a personalized, adaptive, and engaging therapy experience, Rehabotics addresses a critical gap in long-term rehabilitation, where patient motivation is often a key determinant of success. The integration of physical assistance (the exoskeleton) with cognitive engagement (the AR games) represents a significant step beyond conventional therapy methods.
System Components
| Component | Function | Key Innovation |
|---|---|---|
| Soft Glove | Collects data on hand movements and force exertion levels. | Enables real-time, data-driven personalization of therapy. |
| Robotic Exoskeleton Hand | Assists patients with hand movements, particularly opening the hand. | Uses servomotors to physically overcome spasticity. |
| Augmented Reality (AR) Platform | Hosts serious games of varying difficulty for adaptive therapy. | Increases patient engagement and adherence through gamification. |
Benchmarking Emotion AI in a Masked World
The COVID-19 pandemic introduced a new challenge for facial emotion recognition systems: the ubiquity of surgical masks. Researchers tackled this problem head-on by benchmarking the performance of a widely used library, DeepFace, on the FER2013 dataset while individuals wore masks 1 .
The results were striking, revealing a substantial performance decline across all emotions. The emotion of "Disgust" was most affected, suffering a 22.6% reduction in F1-score, while "Surprise" was the least affected, though it still saw a significant 48.7% reduction 1 .
Research Significance
This research is crucial for developing more robust human-machine interfaces that can function accurately in real-world conditions.
Impact of Surgical Masks on Emotion Recognition Accuracy (F1-Score Reduction) 1
Predicting Heart Wellness from Skin Color
In the realm of non-invasive monitoring, a study explored remote photoplethysmography (rPPG) for predicting heart wellness by analyzing skin color variations in video data 1 . The research involved 20 participants from diverse backgrounds, considering factors like gender, skin texture, and age group 1 .
The study evaluated the effectiveness of different rPPG methods using machine learning algorithms. The Random Forest Regression model emerged as particularly accurate, achieving an average mean square error of 3.193 and a coefficient of determination (R²) value of 0.885, indicating a strong predictive model 1 .
Future Applications
This work paves the way for accessible, camera-based heart rate monitoring that could be integrated into smartphones or laptops.
Performance of Machine Learning Models in rPPG Heart Rate Prediction 1
| Machine Learning Model | Average Mean Square Error | Coefficient of Determination (R²) |
|---|---|---|
| Random Forest Regression | 3.193 | 0.885 |
| Lasso Regression | 33.336 | Information missing in source |
The Scientist's Toolkit: Essential Research Reagent Solutions
The breakthroughs presented at ABSET 2023 rely on a sophisticated arsenal of tools and technologies. The following details some of the key "research reagents" and materials that are foundational to this field.
| Tool/Technology | Function in Research | Example Use Case |
|---|---|---|
| Active Shape Models (ASM) | A statistical model of object shape that can deform to fit new images. | Used for the bimodal boundary extraction of cervical vertebrae and spinal canals from CT and MR images 1 . |
| Robotic Exoskeletons | Wearable devices that provide mechanical support and assistance to limbs. | Assists post-stroke patients in performing hand-opening movements during rehabilitation therapy 1 . |
| Serious Games (Augmented Reality) | Game applications designed for a primary purpose other than pure entertainment. | Provides adaptive, engaging motor and cognitive tasks for patients in occupational therapy 1 . |
| Remote Photoplethysmography (rPPG) | A technology that measures physiological signals by detecting subtle skin color changes from a distance. | Enables non-contact heart rate estimation from a standard video feed 1 . |
| Gait Analysis Systems (e.g., Vicon) | Motion capture systems that precisely track body movement in 3D. | Quantifies the effect of an interventional movement program on the hip joint mechanics of a patient with dementia 1 . |
| Machine Learning Algorithms (e.g., Random Forest) | Computational methods that learn patterns from data to make predictions or decisions. | Used to accurately predict heart rate from rPPG data and to benchmark emotion recognition algorithms 1 . |
Conclusion: Engineering a Healthier Tomorrow
The research unveiled at the ABSET 2023 conference paints a vivid picture of a future where technology and biology are inextricably linked for human benefit. From the Rehabotics platform that turns recovery into an engaging journey, to the algorithms that see our heart rate through our skin and our emotions behind our masks, these innovations are not merely incremental improvements. They are paradigm shifts.
Note on Source
This article is based on the proceedings of the Advances in Biomedical Sciences, Engineering and Technology (ABSET) 2023 conference, as published in Engineering Proceedings. The conference was organized by the Department of Biomedical Engineering of the University of West Attica, Greece, and featured research from scientists across 15 countries 1 2 .