Development and prospect of near-infrared spectroscopy-assisted schizophrenia diagnosis based on bibliometrics

Abstract

In this editorial, we comment on the recent article by Fei et al exploring the field of near-infrared spectroscopy (NIRS) research in schizophrenia from a bibliometrics perspective. In recent years, NIRS has shown unique advantages in the auxiliary diagnosis of schizophrenia, and the introduction of bibliometrics has provided a macro perspective for research in this field. Despite the opportunities brought about by these technological developments, remaining challenges require multidisciplinary approach to devise a reliable and accurate diagnosis system for schizophrenia. Nonetheless, NIRS-assisted technology is expected to contribute to the division of methods for early intervention and treatment of schizophrenia.

Key Words: Bibliometrics; Schizophrenia; Near-infrared spectroscopy; Diagnostic technique; Data analysis

Core Tip: This manuscript employs bibliometric analysis to investigate the application of near-infrared spectroscopy (NIRS) in the assisted diagnosis of schizophrenia. NIRS, a non-invasively user-friendly functional brain imaging modality, rapidly provides critical biomarker information via high-throughput data acquisition and advanced signal processing algorithms. The findings indicate deep learning and machine learning techniques, including random forests and support vector machines, enhance the accuracy and robustness of NIRS data analysis. Nonetheless, challenges remain in processing large datasets and integrating heterogeneous data sources. Future advancements in equipment, algorithms, and multi-center collaboration are anticipated to further elevate NIRS's clinical utility and reliability in schizophrenia diagnosis.

INTRODUCTION

Schizophrenia is a serious mental disorder affecting the quality of life of millions of people worldwide. Its complex etiology, diverse clinical manifestations, and individual differences in response to treatment make diagnosis and development of treatment strategies challenging[1]. Near-infrared spectroscopy (NIRS), a noninvasive and easy-to-operate functional brain imaging tool, has shown unique advantages in the auxiliary diagnosis of schizophrenia[2-4]. Combined with high-throughput data acquisition capabilities and advanced signal processing algorithms, NIRS is able to provide critical biomarker information in a short period of time.

The introduction of bibliometrics in recent years has provided a macro perspective for research in this field, revealing research and development trends through big data analysis[5-7]. Through statistical analysis of thousands of literatures related to the role of NIRS in the diagnosis of schizophrenia, key word frequencies and research hotspots as well as international cooperation networks have been identified. Further data mining work shows that the introduction of deep learning technology promotes the accuracy of pattern recognition in data analysis, which means that the analysis of NIRS signals will be more refined and efficient. However, with the processing of massive data and fusion of multisource heterogeneous data, more powerful computational methods need to be developed to improve the generalization ability and diagnostic efficacy of prediction models.

Going forward, the application of NIRS in schizophrenia diagnosis may benefit from further advances in artificial intelligence techniques, in particular, enhancing data processing and understanding complex brain network dynamics. Through refined study designs and large-scale multicenter collaboration, this technology is expected to become an indispensable tool in precision medicine for schizophrenia in the future.

APPLICATION OF BIBLIOMETRICS IN SCHIZOPHRENIA RESEARCH

With the rapid development of statistics and information science and technology, the application of bibliometrics in the medical field has gradually shown its unique research value. Through mining and quantitative analysis of literature data published in major medical journals, researchers can visualize the macro status and trend change in disease research[8-11]. Understanding the research hotspots and trends allows for tracking the development of diagnostic and treatment technologies for various diseases, including schizophrenia[12-14].

In recent years, bibliometrics has been widely used in schizophrenia research, particularly for tracking and analyzing the knowledge structure and development trend in this field[15-19]. Using bibliometric tools and algorithms, researchers are able to systematically evaluate the scientific literature related to schizophrenia. Through co-word and cluster analyses, keyword co-occurrence networks can be constructed to refine the core areas of schizophrenia diagnosis, treatment, and mechanism research.

Although bibliometric analysis provides an innovative perspective for the diagnosis and treatment of schizophrenia, this approach currently faces several challenges, such as improving the coverage and accuracy of data collection, optimizing the model algorithm to improve the accuracy of prediction, and ensuring rational interpretation of the biological significance of complex data.

APPLICATION OF NIRS IN SCHIZOPHRENIA DIAGNOSIS

NIRS is based on molecular vibrational spectroscopy, which relies on the absorption properties of chemical bonds in molecules within the near-infrared region, such as hydrogen bonds, and can be used to detect and analyze biochemical components in biological tissues. The application of this technology in the medical field is increasingly showing its value because of its significant advantages of noninvasiveness and high sensitivity, particularly in the auxiliary diagnosis of neurological and psychiatric diseases, such as schizophrenia, demonstrating its potential as a tool for early diagnosis and disease monitoring. NIRS enables quick monitoring of changes in blood oxygen levels in brain tissue through spectral scans of specific areas of the brain, as well as capturing spectral data from biological samples of patients, such as blood or brain tissue. These data are then analyzed using advanced data processing algorithms, such as principal component analysis and partial least squares regression. Building spectral models of schizophrenia-related biomarkers, combined with machine learning methods like random forest (RF) and support vector machine (SVM), yields innovative ways to improve the accuracy and robustness of diagnostic models, thereby providing physicians with important clues about functional and structural brain abnormalities.

Studies of brain function in patients with schizophrenia have demonstrated differences in spectral absorption characteristics between normal and pathological brain tissue. NIRS is sensitive enough to detect these subtle changes, offering possibilities for early disease detection and monitoring. Current research focuses on the optimization of NIRS parameters, including the accuracy of probe placement, selection of light source wavelength, and improvement of signal processing algorithms, to enhance the sensitivity of detection of schizophrenia-specific biomarkers. Although difficult problems need to be overcome in its practical application, such as motion artifacts and signal variation due to individual differences, the potential of NIRS as an auxiliary diagnostic tool is gradually being recognized by researchers and clinicians globally.

Future directions may include optimizing spectral acquisition devices, improving the robustness of data processing algorithms, and conducting more extensive clinical trials to verify the stability and reliability of this technology in different populations.

TRENDS IN THE DEVELOPMENT OF NIRS-ASSISTED SCHIZOPHRENIA DIAGNOSTIC TOOLS

Bibliometric methods are widely used to evaluate the scientific research productivity and knowledge structure in the field of NIRS-assisted diagnosis of schizophrenia to provide data support for the direction of scientific research. In their recent study, Fei et al[20], published in the World Journal of Psychiatry: Near-infrared spectroscopy in schizophrenia: A bibliometric perspective to explore the current situation and trend of NIRS in schizophrenia from the perspective of bibliometric, explored the current status and trend of NIRS application in schizophrenia from bibliometric perspective. They reported that, with the continuous progress of technology and reduction of cost, NIRS is increasingly accepted and applied by medical institutions and research teams, providing new ideas and methods for the clinical diagnosis and treatment of schizophrenia. Their in-depth analysis revealed a complex pattern of association between molecular spectral features and patients' cognitive function, and indicated directions for future research on auxiliary diagnostic tools.

With the rapid development of computational techniques and spectral analysis methods, NIRS-assisted schizophrenia diagnosis has attracted great attention in the research field. The application of machine learning algorithms to process and analyze spectral data has made a major breakthrough in improving the accuracy and speed of diagnosis. In particular, studies using SVM, RF, and deep learning networks such as short-term memory networks, have found that patients with schizophrenia can be accurately identified by biomarkers in spectral signals. In addition, complex spectral data can be denoised and compressed by feature extraction and dimensional-reduction techniques, such as principal component analysis and linear discriminant analysis, enabling the extraction of key biochemical indicators. The current research trend shows a new focus on the development and application of ensemble learning methods, which can combine the advantages of multiple algorithms to improve the generalization ability and stability of models. Predictive models, combined with patient clinical information and spectral data, are expected to enable personalized medicine and provide a more accurate diagnostic basis for schizophrenia.

Current research focuses on exploring the correlation between NIRS signals and schizophrenia symptoms, developing more accurate data analysis algorithms, and building machine learning-based predictive models. Deep learning algorithms have shown remarkable ability in processing complex NIRS data to identify disease-related biomarkers and improve diagnostic accuracy. Nonetheless, NIRS data processing methods still have certain limitations, such as low signal-to-noise ratio and susceptibility to physiological noise interference, which are the key to current technological advances. Future research should focus on optimizing the spectral acquisition device, improving the generalization ability of the algorithm, and exploring multimodal data fusion strategies, with the aim to improve the accuracy and reliability of NIRS-assisted schizophrenia diagnosis.

Despite the challenges of sample size limitations and algorithm interpretation, continuing to explore accurate diagnostic models based on NIRS technology will provide a strong impetus for the development of medical diagnostic technology in the future.

CONCLUSION

In recent years, bibliometrics-based research on schizophrenia diagnosis has shown an innovative trend of multidisciplinary approaches. Remarkable progress has been made in the field of schizophrenia diagnosis assisted by NIRS[21-23], a noninvasive and easy-to-operate tool with extensive potential for clinical practice application. Cerebral cortex oxygen-dependent signals collected by NIRS can reveal dyshemispherization of cognitive function in patients with schizophrenia[24], which is highly consistent with traditional neuropsychological assessment results. Through combination with statistical learning and pattern recognition methods, such as SVM and RF, the classification accuracy of NIRS data has reached a high level, which is of great significance for the early diagnosis and treatment of schizophrenia. Further, the combination of semi-supervised learning and feature selection techniques can identify biomarkers closely related to the disease from thousands of NIRS features, allowing for a more personalized and precise auxiliary diagnosis of schizophrenia. Despite the limitations in sample size and cross-population validation, the strategy of NIRS combined with machine learning undoubtedly provides an emerging and feasible approach for schizophrenia diagnosis. With the continuous development of algorithm optimization and instrument technology, this methodology is expected to achieve rapid and accurate screening and diagnosis in a wider range of people, and contribute to the division of methods for early intervention and treatment of schizophrenia[25-28]. The combination of high-throughput spectral analysis and deep learning algorithms is expected to greatly improve the diagnostic accuracy and efficiency in the future. Deep learning models, such as convolutional and recurrent neural networks, have demonstrated superiority in processing and recognizing patterns of complex biomarkers' spectral lines, which will further optimize the early monitoring and classification process of diseases. In addition, transfer learning technology can improve the generalization ability of models across different devices by utilizing spectral information from existing databases, thereby enhancing their applicability in clinical practice. Quality control of spectral data and new preprocessing techniques will solve the problem of spectral noise and baseline drift, reducing diagnostic errors. Despite the unprecedented opportunities brought about by these technological developments, challenges remain, such as standardization of spectral acquisition conditions, transparency of data processing methods, and model interpretability. Therefore, multidisciplinary research will play a key role in building a reliable and accurate diagnosis system for schizophrenia. Overall, the insights based on bibliometrics will indicate the research direction for the application of NIRS in the diagnosis of schizophrenia, and guide future technological innovation and clinical transformation.