Trilateral AI-Cog JP+DE+FR

Project Title

"AI for Aging Societies: From Basic Concepts to Practical Tools for AI-Facilitated Cognitive Training" 

Japanese Team Members

• Japanese team PI: Dr. Mihoko OTAKE-MATSUURA, RIKEN AIP

• Research scientist: Dr. Tomasz M. RUTKOWSKI, RIKEN AIP

• Postdoctoral researcher: Dr. Kornkanok TRIPANPITAK, RIKEN AIP

• Postdoctoral researcher:  Dr. Alexandra WOLF, RIKEN AIP

German Team Members

• German team PI: Prof. Tonio BALL, Neuromedical AI Lab, Albert-Ludwigs-University Freiburg

• Doctoral candiate: Maryna KAPITANOVA, Neuromedical AI Lab, Albert-Ludwigs-University Freiburg

French Team Members

• French team PI: Prof. Thomas MOREAU, INRIA

• Doctoral candiate: Apolline MELLOT, INRIA

AI for Aging Societies - Executive Summary 

People live longer than ever, most now living beyond their sixties. Aging societies bring significant social, economic, and healthcare challenges. Japan (#1), France (#3), and Germany (#4) are among the top five countries worldwide, with the highest proportion of people over 65. Health conditions in older age include depression, dementia, and long-term disability after stroke. About 50 million people worldwide suffer from dementia, with almost 10 million new cases yearly. According to WHO, there is a new case of dementia every three seconds globally.

Aging populations and age-related brain disorders are a global concern, particularly in Japan, France, and Germany. This project aimed to improve healthy aging by utilizing artificial intelligence (AI) to study machine-learning-driven biomarkers, developing dedicated ML methods for the human brain, and creating an open-source reference software package. The project's scientific objectives included optimizing understandable information about the brain's functional state, identifying biomarkers for cognitive impairments, and guiding AI-facilitated cognitive training.

Despite the challenges of conducting EEG experiments and cognitive interventions during COVID-19 lockdowns in Japan, three research groups from Germany, France, and Japan have redirected their focus towards developing AI-related neuro-biomarkers from pre-existing datasets and exploring the ethical and societal implications of AI in the context of aging. This project has brought together complementary expertise from these groups, including technologies for interventions, EEG measurements, and statistical machine learning algorithms for data analysis. 

The Japanese team focused on analyzing data from pre-pandemic experiments conducted by the CBAT Team at RIKEN AIP. They utilized machine learning approaches based on supervised learning principles with features drawn from EEG signal complexity (detrended fluctuation analysis [4-12], network neuroscience [1], and topological data analysis [14]). The experimental paradigms developed and conducted by the CBAT team resulted in encouraging results on a still limited sample size due to pandemic restrictions, reaching the middle ninety percent. The latest results also allowed the team to employ unsupervised learning approaches [16], which was one of the primary targets of the trilateral project. The Japanese team also started a new experimental session in Japan and, in collaboration with Nicolaus Copernicus University in Poland, collected a new database of combined EEG, eye-tracking (ET), and fNIRS datasets for the no-cost extension of the project. The Japanese team conducted a series of public outreach events to promote the use of neurotechnology for detecting dementia biomarkers. These events included online BCI Spring Schools, which had significant participation from people in various countries worldwide. The 2023 edition had 15,787 attendees from 113 countries (https://www.gtec.at/spring-school-2023/), the 2022 edition had 5,309 attendees from 107 countries (https://www.gtec.at/spring-school-2022/), and the 2021 edition had 4,025 attendees (https://www.gtec.at/spring-school-2021/). Additionally, they co-organized the Neuro-diversity Change Tomorrow Takeshiba 2023 event, with over 100 participants visiting their EEG demonstration booth (https://www.change-tomorrow-takeshiba.com/). 

The German team conducted a broad study to attack a question of the brain-age prediction from EEG for a possible neuro-biomarker. The study [2] achieved state-of-the-art results in age decoding, with a mean absolute error of 6.6 years. The researchers found that the model underestimated the age of both non-pathological and pathological subjects by -1 and -5 years, respectively. However, there were no significant differences in the average brain age gap between the two groups, whether they underwent single or repeated examinations. These results contradict previous studies that supported the brain age hypothesis. Therefore, the brain age gap biomarker cannot be used to indicate pathological EEG in different datasets. A group of researchers from Germany conducted a study to investigate the relationship between the acquisition or recovery of pathology and the brain age gap in different datasets. The results of their study indicate that the trait hypothesis is more likely than the state hypothesis for brain age estimates derived from EEG, as no significant change was observed in the brain age gap. As a result, the initial publication [3] by the French and German teams focused on training supervised machine learning models to estimate subject age from brainwaves, which was not helpful for the dementia neuro-biomarker application. The French team participated in the BIOMAG 2022 Competition on Dementia Screening and emerged as the winner. Their approach to feature extraction based on power densities and covariance tangent vectors was quite impressive. It led to excellent results, placing them at the top of the MEG dataset in the challenge (https://biomag2020.org/awards/data-analysis-competitions/). 

Newly Added/Modified Research Plan 

The ongoing COVID-19 pandemic has significantly impacted our research project, as well as many others across the globe. Due to the restrictions imposed to curb the virus spread, we had to reevaluate our initial research plans. The pandemic has presented unprecedented challenges, particularly in collecting experimental data. 

In response to the pandemic, we shifted our focus from experimental data collection to processing existing datasets and developing novel AI methods to support our research goals. This shift has demonstrated our ability to adapt and pivot our research plans to accommodate the changing circumstances. Although the pandemic has been a challenging time for many, it has also highlighted the importance of being flexible and creative in our approach to research. The pandemic has also emphasized the importance of collaboration and sharing data and resources to support research efforts.

To support our research goals, we developed novel feature extraction, supervised and unsupervised machine learning, online research dissemination, and public outreach approaches by participating in global online events and local educational activities. We plan to extend our project for an additional year, at no cost, to finally collect our own datasets and test the developed AI methods.

Research Achievements

Remarkable Research Achievements

• Summary Description of the first result: A team of Japanese researchers and colleagues from Nicolaus Copernicus University in Poland studied emotional responses and cognitive decline in a cross-cultural setting. The study was conducted in Poland, where experimental restrictions due to the pandemic were less severe. This enabled the team to reproduce pre-pandemic emotional faces and interior paradigms. The study utilized EEG complexity features and topological data analysis measures to classify MCI and regress MoCA scores. The results have been published in [1] and are currently under review in [2]. These outcomes have demonstrated that EEG-based neuro-biomarkers could be used as feasible predictors of age-related cognitive decline. To share new datasets among all project members without previous ethical restrictions, subsequent more extensive studies are planned during the project no-cost extension.

• Summary Description of the second result: The French team contributed by winning the BIOMAG 2022 Competition on Dementia Screening. They demonstrated a successful application of MEG power densities and covariance tangent vectors to predict the onset of dementia, which was quite impressive. This impressive and meaningful outcome was a clever workaround for the experimental restrictions caused by the pandemic, which significantly impacted our trilateral project. The French colleagues have shown that applying AI tools to publicly available datasets can lead to a competitive and successful approach, resulting in a top score

• Summary Description of the third result: In a study published in [3], French and German researchers demonstrated that they could predict a person’s brain age with a small margin of error using brainwave data. However, a subsequent thorough investigation led by German colleagues and reported in [2], showed that a brain age gap biomarker was unreliable for indicating pathological EEG across different datasets. This finding significantly impacted the M/EEG scientific community, as it demonstrated that using AI techniques to predict a person’s age from brainwave recordings would not lead to a diagnosis of dementia or brain pathological atrophy. With their negative findings, the German team made a solid and impactful contribution to neurotechnology and neuroscience in general.

Achievements that significantly contribute to science and technology innovation

•   Summary Description for the first result: Japanese researchers and colleagues from Nicolaus Copernicus University in Poland conducted a cross-cultural study on emotional responses and cognitive decline. The study utilized advanced techniques such as EEG complexity features and topological data analysis measures to classify MCI and regress MoCA scores. The results suggest that EEG-based neuro-biomarkers could predict age-related cognitive decline, which could have significant implications for neurology and diagnostic tools. The team's ability to reproduce pre-pandemic emotional faces and interior paradigms when experimental restrictions were severe and their commitment to open science and collaboration demonstrate their innovative approach.

• Summary Description for the second result:  The win of the French team in the BIOMAG 2022 Competition for Dementia Screening was an impressive achievement. Despite the restrictions caused by the pandemic, they cleverly adapted by utilizing an online database provided by the competition organizers. They could make accurate predictions by using MEG power densities and covariance tangent vectors in a novel way. Their use of AI tools on publicly available data showcases their innovative approach. Their top score is a testament to their expertise in this field.

• Summary Description for the third result: A recent study by researchers from France and Germany has made significant progress in predicting a person's brain age using brainwave data. However, further investigation led by German colleagues revealed that relying on a brain age gap biomarker was unreliable for indicating pathological EEG across different datasets. Despite this setback, the German team's findings have contributed significantly to neurotechnology and neuroscience, highlighting the importance of continued research and innovation. Their negative results have paved the way for further advancements and have helped researchers better understand the limitations of using AI techniques to predict a person's age from brainwave recordings, ultimately leading to a better understanding of dementia and brain pathological atrophy.

Representative Papers

• Rutkowski TM, Abe MS, Komendzinski T, Sugimoto H, Narebski S, Otake-Matsuura M. Machine Learning Approach for Early Onset Dementia Neurobiomarker Using EEG Network Topology Features. Frontiers in Human Neuroscience;17:262.

Summary Description: The Japanese team’s research paper published in Frontiers in Human Neuroscience [1] focused on utilizing machine learning algorithms to improve the well-being of individuals with disabilities in the "AI for social good" domain. We explored cognitive-behavioral intervention management and digital non-pharmacological therapies for early-onset dementia neuro-biomarkers. Our pilot study in Poland provided findings for cognitive decline prediction. The proposed three experimental tasks in the current pilot study showcased the vital role of artificial intelligence in predicting early-onset dementia in older adults.

• Gemein LA, Schirrmeister RT, Boedecker J, Ball T. Brain Age Revisited: Investigating the State vs. Trait Hypotheses of EEG-derived Brain-Age Dynamics with Deep Learning. arXiv preprint arXiv:231007029. 2023.

Summary Description: The difference between biological and chronological brain age has been disputed. A recent study concluded that the difference is stable and not subject to significant changes in neurological disorders. However, a German team using clinical EEG recordings found evidence to the contrary [2]. Brain age is a trait that remains stable over time. Further research on brain age using different imaging modalities is required. An expanded dataset is needed for more nuanced investigations of brain age. This dataset should ideally include non-binary pathology scores compiled by multiple reliable experts and accurate, anonymized age information.

• Engemann DA, Mellot A, Hoechenberger R, Banville H, Sabbagh D, Gemein L, et al. A reusable benchmark of brain-age prediction from M/EEG resting-state signals. Neuroimage. 2022;262:119521.

Summary Description: Machine learning on brain images can create measurable indicators for individual aging, known as brain age. MEG and EEG can evaluate brain health on a larger scale, but further research is needed to handle the complexity and diversity of M/EEG signals in various situations. French and German researchers developed reusable benchmarks for machine learning approaches in brain age modeling based on M/EEG adaptations of the BIDS standard [3]. The best performance was achieved by pipelines and architectures incorporating spatially aware representations of the M/EEG signals. These benchmarks come with open-source software and high-level Python scripts, making them reusable for modeling specific cognitive variables or clinical endpoints.  

Open Software Packages Developed During the Project

• Covariance Data Frames for Predictive M/EEG Pipelines - https://github.com/coffeine-labs/coffeine - a library for building biomedical prediction models with M/EEG signals. It uses M/EEG covariance matrices as signal representations and integrates with PyRiemann, scikit-learn, and MNE.

• M/EEG brain age prediction benchmark paper - https://github.com/meeg-ml-benchmarks/brain-age-benchmark-paper - a repository containing code, tools, and resources for a reusable benchmark of brain-age prediction from M/EEG resting-state signals has been developed [3]. It involves the reuse and improvements of braindecode that was released multiple times during our grant cycle. 

Activities During the Research Period 

Workshops and Outreach Activities 

Despite pandemic-related travel and meeting restrictions, we successfully hosted several outreach events. The BCI Spring Schools were held online and had significant participation from people worldwide. In-person events included The First Workshop on Complex Systems Science & Health Neuroscience in Kyoto, Japan, and the RIKEN AIP & NCU Workshops 2023 in Toruń, Poland. We also co-organized a public research outreach event called Neuro-Diversity Change Tomorrow Takeshiba 2023. 

The BCI Spring Schools

The BCI Spring Schools are a series of annual academic and educational events that bring together experts, researchers, and enthusiasts from all over the world to share their knowledge and insights on Brain-Computer Interfaces (BCI). These events are free and open to everyone who wants to learn about the latest developments in this field, regardless of their background or experience. Over the years, the BCI Spring Schools have become increasingly popular, attracting diverse attendees from different countries, cultures, and disciplines. The most recent edition of the event was a huge success, with an impressive 15,787 attendees from 113 different countries. On the first days of the events, our team had the opportunity to present the latest developments in dementia neuro-biomarkers from RIKEN AIP, a leading research institute in Japan. The presentations were well-received and helped showcase our team's activities and opportunities on a global scale. In addition to the dementia neuro-biomarkers, we also shared the latest advancements in the passive BCI modality, which was the primary focus of our team in the Trilateral Cog-AI project. This project aims to develop new and innovative ways to improve the lives of people with cognitive disabilities using BCI technology. Our team is proud to have been a part of the BCI Spring Schools, and we look forward to continuing our work in this exciting and rapidly evolving field.

Academic Workshops

Our recent in-person events focused on presenting the latest research in machine learning approaches applied to brainwave data (such as EEG, fNIRS, fMRI) or behavioral data (including eye-tracking, EMG, EOG, response time) in the field of neurotechnology. These events were organized to bring together experts worldwide to discuss the latest developments and applications related to the health and well-being of the elderly. One of these events was hosted at Doshisha University in Kyoto, Japan, and another was the SCIS&ISIS2022 international conference held in Ise-Shima, Japan. The third event, the RIKEN AIP & NCU Workshops 2023, was held at Nicolaus Copernicus University in Toruń, Poland. During these events, researchers presented papers on various topics, including recent contributions to early-onset neuro-biomarker development for dementia, cognitive load monitoring, and advances in comprehensive brain-computer/machine interfaces (BCI/BMI). A special session also focused on contemporary machine learning approaches, including shallow or deep learning, information processing, and experimental contributions. Overall, these events provided a platform for experts to share their knowledge, insights, and research findings, which will contribute to the advancement of neurotechnology applications and ultimately improve the quality of life for the elderly.

Development Activities that Contribute to Society

Our team has been working on a cutting-edge system to monitor elderly individuals with dementia. The system uses multiple data collection methods like EEG (electroencephalography), fNIRS (functional near-infrared spectroscopy), and eye-tracking. These methods allow us to collect data on the mental processes of elderly individuals, such as attention, memory, and decision-making, which can be affected by dementia. The system is designed to be used remotely in the homes of elderly individuals to help monitor their cognitive well-being. This is especially important during the pandemic when elderly individuals are at a higher risk of contracting the virus and may be isolated from their loved ones. The remote monitoring system allows us to keep track of their cognitive health and intervene as needed. Due to the pandemic, it has been challenging to collect data in laboratories. Therefore, we have developed a prototype that allows for data collection outside of research facilities. The prototype uses affordable and wearable EEG devices that are easy to use and non-invasive. The system is managed centrally, which enables real-time monitoring of participants and easy analysis of experimental progress. Our goal is to use this system to estimate cognitive function, monitor interventions, and detect the early onset of dementia in elderly individuals. By doing so, we can improve remote healthcare and promote well-being among the elderly. Additionally, this system provides an objective way to collect neurophysiological data during pandemics when hygiene restrictions are in place. Overall, our system has the potential to revolutionize remote healthcare for the elderly and improve their quality of life.

Future Plan of Research

As part of the trilateral project, which aims to develop neuro-biomarkers for the early detection of dementia, the future research plan includes a collection of more EEG, fNIRS, and eye-tracking data from elderly participants in Japan and Poland. This will enable us to validate further the neuro-biomarker candidates developed thus far and identify any additional biomarkers that may aid in detecting dementia. Dementia is a progressive disease that affects cognitive function, memory, and behavior. It is estimated that 50 million people worldwide are living with dementia, and this number is projected to triple by 2050. Early detection of dementia is crucial for the effective management of the disease, as it allows for early intervention and treatment. The trilateral project, which involved researchers from Japan, France, and Germany, aimed to develop neuro-biomarkers that could detect early signs of dementia before the onset of clinical symptoms. The project has already identified several promising neuro-biomarker candidates showing significant brain activity differences between individuals with and without dementia. To further validate these neuro-biomarker candidates, we plan to collect more EEG, fNIRS, and eye-tracking data from elderly participants in Japan and Poland. We will use advanced statistical methods to analyze this data and identify any additional biomarkers that may aid in detecting dementia. In addition to validating the neuro-biomarker candidates, we intend to publish open-source software and test datasets. This will enable other researchers to replicate our findings and contribute to developing new neuro-biomarkers for dementia. Unfortunately, the COVID-19 pandemic has disrupted our research plans, but we remain committed to our goals. We will take all necessary precautions to ensure the safety of our participants and researchers while collecting data. We believe that the trilateral project has the potential to make a significant contribution to the early detection and management of dementia, and we are excited to continue our research in this field.

Future Collaboration and Prospects for Research Results 

Aging is a natural process that affects all humans. Age-related cognitive decline is a significant public health concern that impacts the quality of life of elderly individuals and their families. Dementia is a common form of age-related cognitive decline affecting millions worldwide. Early diagnosis and timely intervention can significantly improve the outcome of age-related cognitive decline and prevent the onset of dementia. Biomarkers are becoming increasingly important in providing early diagnosis and predicting the progression of age-related cognitive decline. However, validating biomarker candidates is challenging and requires collaboration among researchers from different countries and cultures. This essay discusses the importance of a multinational and multimodal approach for validating biomarker candidates for age-related cognitive decline. 

Age-related cognitive decline is a global issue that affects individuals from different cultures and backgrounds. Therefore, validating biomarker candidates requires collaboration among researchers from different countries and cultures. A multinational approach ensures that the biomarker candidates are validated using data from diverse populations, improving the results' accuracy and reliability. Additionally, a multimodal approach provides a more comprehensive understanding of age-related cognitive decline by combining different techniques to validate the biomarker candidates. EEG, fNIRS, and eye-tracking techniques provide a multimodal approach to validate the biomarker candidates and improve the accuracy of the results.

The proposed methodology involves a collaborative effort among researchers from different countries and cultures. The current Japanese, German, and French project members plan to collaborate and extend their collaboration to Poland, resulting in a multinational and multimodal database of biomarker candidates for age-related cognitive decline. The biomarker candidates will be validated using experiments conducted with elderly individuals from various backgrounds and cultures. The experiments will include EEG, fNIRS, and eye-tracking techniques to provide a multimodal approach to validate the biomarker candidates. Additionally, collaborations will be initiated with researchers from Dutch and American institutions to expand the data collection. 

The expected outcomes of this research are as follows:  a multinational and multimodal database of biomarker candidates for age-related cognitive decline; a validated set of biomarker candidates for age-related cognitive decline; a better understanding of how different cultures and backgrounds affect age-related cognitive decline and the validation of biomarker candidates. 

The implications of the research are significant. The validation of biomarker candidates using a multinational and multimodal approach will provide a better understanding of age-related cognitive decline and the factors that contribute to it. Additionally, the validated set of biomarker candidates will improve the accuracy and reliability of early diagnosis and prediction of the progression of age-related cognitive decline. This research will contribute significantly to the early detection and prevention of the onset of dementia, which is a significant public health concern. 

In conclusion, validating biomarker candidates for age-related cognitive decline using a multinational and multimodal approach is crucial. The collaboration among researchers from different countries and cultures will provide a better understanding of how different cultures and backgrounds affect age-related cognitive decline and the validation of biomarker candidates. The validated set of biomarker candidates will improve the accuracy and reliability of early diagnosis and prediction of the progression of age-related cognitive decline. This research will contribute significantly to the early detection and prevention of the onset of dementia, which is a significant public health concern. 

References

1. Rutkowski TM, Abe MS, Komendzinski T, Sugimoto H, Narebski S, Otake-Matsuura M. Machine Learning Approach for Early Onset Dementia Neurobiomarker Using EEG Network Topology Features. Frontiers in Human Neuroscience;17:262.

2. Gemein LA, Schirrmeister RT, Boedecker J, Ball T. Brain Age Revisited: Investigating the State vs. Trait Hypotheses of EEG-derived Brain-Age Dynamics with Deep Learning. arXiv preprint arXiv:231007029. 2023.

3. Engemann DA, Mellot A, H ̈ochenberger R, Banville H, Sabbagh D, Gemein L, et al. A reusable benchmark of brain-age prediction from M/EEG resting-state signals. Neuroimage. 2022;262:119521.Rutkowski TM, Koculak M, Abe MS, Otake-Matsuura M. Brain Correlates of Task–Load and Dementia Elucidation with Tensor Machine Learning Using Oddball BCI Paradigm. In: ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP); 2019. p. 8578–8582.

4. Rutkowski TM, Komendzinski T, Otake-Matsuura M. Mild Cognitive Impairment Prediction and Cognitive Score Regression in the Elderly using EEG Topological Data Analysis and Machine Learning with Awareness Assessed in Affective Reminiscent Paradigm. In: Frontiers in Aging Neuroscience; 2023. p. (submitted, in review).

5. Rutkowski TM, Abe MS, Koculak M, Otake-Matsuura M. Cognitive Assessment Estimation from Behavioral Responses in Emotional Faces Evaluation Task - AI Regression Approach for Dementia Onset Prediction in Aging Societies. In: NeurIPS Joint Workshop on AI for Social Good - Track 1 - Producing Good Outcomes. Vancouver, Canada; 2019. p. 1-4.

6. Rutkowski TM, Zhao Q, Abe MS, Otake M. AI Neurotechnology for Aging Societies – Task-load and Dementia EEG Digital Biomarker Development Using Information Geometry Machine Learning Methods. In: AI for Social Good Workshop at the Neural Information Processing Systems (NeurIPS = formerly NIPS) 2018. Montreal, Canada; 2018. p. 1-4. 

7. Rutkowski TM, Abe MS, Koculak M, Otake-Matsuura M. Classifying Mild Cognitive Impairment from Behavioral Responses in Emotional Arousal and Valence Evaluation Task - AI Approach for Early Dementia Biomarker in Aging Societies -. In: 42nd Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). IEEE Engineering in Medicine and Biology Society. Montreal, Canada: IEEE Press; 2020. p. 5537-5543.

8. Rutkowski TM, Abe MS, Otake-Matsuura M. EEG and fNIRS Biomarkers of Dementia Prediction and Monitoring. In: The 42nd Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). vol. Mini-Symposia: Brain-computer Interfaces for Neuromodulation, Language, Decoding, high-gamma Mapping, and Control. IEEE Engineering in Medicine and Biology Society. Montreal, Canada: IEEE Press; 2020. p. 6.

9. Rutkowski TM, Abe MS, Otake-Matsuura M. Passive BCI for Dementia Onset Detection and Cognitive Intervention Monitoring. In: International Symposium on Artificial Intelligence and Brain Science 2020. Tokyo, Japan: The University of Tokyo; 2020. p. 47.

10. Rutkowski TM. Passive BCI Paradigm to Elucidate Early Dementia Onset. In: The 43rd Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). IEEE Engineering in Medicine and Biology Society. Virtual: IEEE Press; 2021. p. 5403.

11. Rutkowski TM, Abe MS, Otake-Matsuura M. Neurotechnology and AI Approach for Early Dementia Onset Biomarker from EEG in Emotional Stimulus Evaluation Task. In: The 43rd Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). IEEE Engineering in Medicine and Biology Society. Virtual: IEEE Press; 2021. p. 6682-6685.

12. Rutkowski TM, Abe MS, Komendzinski T, Otake-Matsuura M. Older Adult Mild Cognitive Impairment Prediction from Multiscale Entropy EEG Patterns in Reminiscent Interior Image Working Memory Paradigm. In: The 43rd Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). IEEE Engineering in Medicine and Biology Society. Virtual: IEEE Press; 2021. p. 6352-6355.

13. Rutkowski TM, Abe MS, Tokunaga S, Komendzinski T, Otake-Matsuura M. Dementia Digital Neuro-biomarker Study from Theta-band EEG Fluctuation Analysis in Facial and Emotional Identification Short-term Memory Oddball Paradigm. In: 2022 44th Annual International Conference of the IEEE Engineering in Medicine Biology Society (EMBC). Glasgow, UK: IEEE Press; 2022. p. 4129-32.

14. Rutkowski TM, Abe MS, Tokunaga S, Sugimoto H, Komendzinski T, Otake-Matsuura M. Passive BCI Oddball Paradigm for Dementia Digital Neuro-biomarker Elucidation from Attended and Inhibited ERPs Utilizing Information Geometry Classification Approaches. In: 2022 IEEE International Conference on Systems, Man and Cybernetics (SMC). IEEE Systems, Man and Cybernetics. Prag, Czechia: IEEE Press; 2022. p. 2654-2659

15. Rutkowski TM, Narebski S, Bekier P, Komendzinski T, Sugimoto H, Otake-Matsuura M. Cross-cultural Evaluation of Dementia Passive BCI Neuro-biomarker Candidate. In: Proceedings of the 2022 Joint 12th International Conference on Soft Computing and Intelligent Systems and 23rd International Symposium on Advanced Intelligent Systems (SCIS & ISIS). IEEE Press; 2022. p. 1-5.

16. Rutkowski TM, Abe MS, Sugimoto H, Otake-Matsuura M. Mild Cognitive Impairment Prediction with Machine Learning and Topological Data Analysis Applied to EEG Time-series in Facial Emotion Oddball Paradigm - Objective Neurobiomarker for Early Onset Dementia Prediction Approach -. In: 2023 45th Annual International Conference of the IEEE Engineering in Medicine Biology Society (EMBC). Sydney, Australia: IEEE Press; 2023. p. (accepted, presented, in press).

17. Kodama T, Rutkowski TM, Komendzinski T, Ito YM, Otake-Matsuura M. Multi-modal EEG/fNIRS/ET Data Collection and Cloud Storage System Prototype for the Elderly Home-based Dementia Monitoring. In: 2023 45th Annual International Conference of the IEEE Engineering in Medicine Biology Society (EMBC). Sydney, Australia: IEEE Press; 2023. p. (accepted, in press).

18. Mellot A, Collas A, Rodrigues PLC, Engemann D, Gramfort A. Harmonizing and aligning M/EEG datasets with covariance-based techniques to enhance predictive regression modeling. bioRxiv. 2023.

19. Rutkowski TM, Komendzinski T, Otake-Matsuura M. Mild Cognitive Impairment Prediction and Cognitive Score Regression in the Elderly using EEG Topological Data Analysis and Machine Learning with Awareness Assessed in Affective Reminiscent Paradigm. Frontiers in Aging Neuroscience;15:1294139.