Transformer-based deep neural network language models for Alzheimer's disease risk assessment from targeted speech

Linguistic-based Mild Cognitive Impairment detection using Informative Loss

This paper presents a deep learning method using Natural Language Processing (NLP) techniques, to distinguish between Mild Cognitive Impairment (MCI) and Normal Cognitive (NC) conditions in older adults. We propose a framework that analyzes transcripts generated from video interviews collected within the I-CONECT study project, a randomized controlled trial aimed at improving cognitive functions through video chats. Our proposed NLP framework consists of two Transformer-based modules, namely Sentence Embedding (SE) and Sentence Cross Attention (SCA). First, the SE module captures contextual relationships between words within each sentence. Subsequently, the SCA module extracts temporal features from a sequence of sentences. This feature is then used by a Multi-Layer Perceptron (MLP) for the classification of subjects into MCI or NC. To build a robust model, we propose a novel loss function, called InfoLoss, that considers the reduction in entropy by observing each sequence of sentences to ultimately enhance the classification accuracy. The results of our comprehensive model evaluation using the I-CONECT dataset show that our framework can distinguish between MCI and NC with an average area under the curve of 84.75%.

Natural language processing techniques for studying language in pathological ageing: A scoping review

BACKGROUND

In the past few years there has been a growing interest in the employment of verbal productions as digital biomarkers, namely objective, quantifiable behavioural data that can be collected and measured by means of digital devices, allowing for a low-cost pathology detection, classification and monitoring. Numerous research papers have been published on the automatic detection of subtle verbal alteration, starting from written texts, raw speech recordings and transcripts, and such linguistic analysis has been singled out as a cost-effective method for diagnosing dementia and other medical conditions common among elderly patients (e.g., cognitive dysfunctions associated with metabolic disorders, dysarthria).

AIMS

To provide a critical appraisal and synthesis of evidence concerning the application of natural language processing (NLP) techniques for clinical purposes in the geriatric population. In particular, we discuss the state of the art on studying language in healthy and pathological ageing, focusing on the latest research efforts to build non-intrusive language-based tools for the early identification of cognitive frailty due to dementia. We also discuss some challenges and open problems raised by this approach.

METHODS & PROCEDURES

We performed a scoping review to examine emerging evidence about this novel domain. Potentially relevant studies published up to November 2021 were identified from the databases of MEDLINE, Cochrane and Web of Science. We also browsed the proceedings of leading international conferences (e.g., ACL, COLING, Interspeech, LREC) from 2017 to 2021, and checked the reference lists of relevant studies and reviews.

MAIN CONTRIBUTION

The paper provides an introductory, but complete, overview of the application of NLP techniques for studying language disruption due to dementia. We also suggest that this technique can be fruitfully applied to other medical conditions (e.g., cognitive dysfunctions associated with dysarthria, cerebrovascular disease and mood disorders).

CONCLUSIONS & IMPLICATIONS

Despite several critical points need to be addressed by the scientific community, a growing body of empirical evidence shows that NLP techniques can represent a promising tool for studying language changes in pathological aging, with a high potential to lead a significant shift in clinical practice.

WHAT THIS PAPER ADDS

What is already known on this subject Speech and languages abilities change due to non-pathological neurocognitive ageing and neurodegenerative processes. These subtle verbal modifications can be measured through NLP techniques and used as biomarkers for screening/diagnostic purposes in the geriatric population (i.e., digital linguistic biomarkers-DLBs). What this paper adds to existing knowledge The review shows that DLBs can represent a promising clinical tool, with a high potential to spark a major shift to dementia assessment in the elderly. Some challenges and open problems are also discussed. What are the potential or actual clinical implications of this work? This methodological review represents a starting point for clinicians approaching the DLB research field for studying language in healthy and pathological ageing. It summarizes the state of the art and future research directions of this novel approach.

Speech and language processing with deep learning for dementia diagnosis: A systematic review

Dementia is a progressive neurodegenerative disease that burdens the person living with the disease, their families, and medical and social services. Timely diagnosis of dementia could be followed by introducing interventions that may slow down its progression or reduce its burdens. However, the diagnostic process of dementia is often complex and resource intensive. Access to diagnostic services is also an issue in low and middle-income countries. The abundance and easy accessibility of speech and language data have created new possibilities for utilizing Deep Learning (DL) technologies to be part of the dementia diagnostic process. This systematic review included studies published between 2012-2022 that utilized such technologies to aid in diagnosing dementia. We identified 72 studies using the PRISMA 2020 protocol, extracted and analyzed data from these studies and reported the related DL technologies. We found these technologies effectively differentiated between healthy individuals and those with a dementia diagnosis, highlighting their potential in the diagnosis of dementia. This systematic review provides insights into the contributions of DL-based speech and language techniques to support the dementia diagnostic process. It also offers an understanding of the advancements made in this field thus far and highlights some challenges that still need to be addressed.

Detection and localization of caries and hypomineralization on dental photographs with a vision transformer model

Caries and molar-incisor hypomineralization (MIH) are among the most prevalent diseases worldwide and need to be reliably diagnosed. The use of dental photographs and artificial intelligence (AI) methods may potentially contribute to realizing accurate and automated diagnostic visual examinations in the future. Therefore, the present study aimed to develop an AI-based algorithm that can detect, classify and localize caries and MIH. This study included an image set of 18,179 anonymous photographs. Pixelwise image labeling was achieved by trained and calibrated annotators using the Computer Vision Annotation Tool (CVAT). All annotations were made according to standard methods and were independently checked by an experienced dentist. The entire image set was divided into training (N = 16,679), validation (N = 500) and test sets (N = 1000). The AI-based algorithm was trained and finetuned over 250 epochs by using image augmentation and adapting a vision transformer network (SegFormer-B5). Statistics included the determination of the intersection over union (IoU), average precision (AP) and accuracy (ACC). The overall diagnostic performance in terms of IoU, AP and ACC were 0.959, 0.977 and 0.978 for the finetuned model, respectively. The corresponding data for the most relevant caries classes of non-cavitations (0.630, 0.813 and 0.990) and dentin cavities (0.692, 0.830, and 0.997) were found to be high. MIH-related demarcated opacity (0.672, 0.827, and 0.993) and atypical restoration (0.829, 0.902, and 0.999) showed similar results. Here, we report that the model achieves excellent precision for pixelwise detection and localization of caries and MIH. Nevertheless, the model needs to be further improved and externally validated.

A robust deep learning workflow to predict CD8 + T-cell epitopes

BACKGROUND

T-cells play a crucial role in the adaptive immune system by triggering responses against cancer cells and pathogens, while maintaining tolerance against self-antigens, which has sparked interest in the development of various T-cell-focused immunotherapies. However, the identification of antigens recognised by T-cells is low-throughput and laborious. To overcome some of these limitations, computational methods for predicting CD8 + T-cell epitopes have emerged. Despite recent developments, most immunogenicity algorithms struggle to learn features of peptide immunogenicity from small datasets, suffer from HLA bias and are unable to reliably predict pathology-specific CD8 + T-cell epitopes.

METHODS

We developed TRAP (T-cell recognition potential of HLA-I presented peptides), a robust deep learning workflow for predicting CD8 + T-cell epitopes from MHC-I presented pathogenic and self-peptides. TRAP uses transfer learning, deep learning architecture and MHC binding information to make context-specific predictions of CD8 + T-cell epitopes. TRAP also detects low-confidence predictions for peptides that differ significantly from those in the training datasets to abstain from making incorrect predictions. To estimate the immunogenicity of pathogenic peptides with low-confidence predictions, we further developed a novel metric, RSAT (relative similarity to autoantigens and tumour-associated antigens), as a complementary to 'dissimilarity to self' from cancer studies.

RESULTS

TRAP was used to identify epitopes from glioblastoma patients as well as SARS-CoV-2 peptides, and it outperformed other algorithms in both cancer and pathogenic settings. TRAP was especially effective at extracting immunogenicity-associated properties from restricted data of emerging pathogens and translating them onto related species, as well as minimising the loss of likely epitopes in imbalanced datasets. We also demonstrated that the novel metric termed RSAT was able to estimate immunogenic of pathogenic peptides of various lengths and species. TRAP implementation is available at: https://github.com/ChloeHJ/TRAP .

CONCLUSIONS

This study presents a novel computational workflow for accurately predicting CD8 + T-cell epitopes to foster a better understanding of antigen-specific T-cell response and the development of effective clinical therapeutics.

Automated Classification of Cognitive Decline and Probable Alzheimer's Dementia Across Multiple Speech and Language Domains

BACKGROUND

Decline in language has emerged as a new potential biomarker for the early detection of Alzheimer's disease (AD). It remains unclear how sensitive language measures are across different tasks, language domains, and languages, and to what extent changes can be reliably detected in early stages such as subjective cognitive decline (SCD) and mild cognitive impairment (MCI).

METHOD

Using a scene construction task for speech elicitation in a new Spanish/Catalan speaking cohort (N = 119), we automatically extracted features across seven domains, three acoustic (spectral, cepstral, and voice quality), one prosodic, and three from text (morpholexical, semantic, and syntactic). They were forwarded to a random forest classifier to evaluate the discriminability of participants with probable AD dementia, amnestic and nonamnestic MCI, SCD, and cognitively healthy controls. Repeated-measures analyses of variance and paired-samples Wilcoxon signed-ranks test were used to assess whether and how performance differs significantly across groups and linguistic domains.

RESULTS

The performance scores of the machine learning classifier were generally satisfactorily high, with the highest scores over .9. Model performance was significantly different for linguistic domains (p < .001), and speech versus text (p = .043), with speech features outperforming textual features, and voice quality performing best. High diagnostic classification accuracies were seen even within both cognitively healthy (controls vs. SCD) and MCI (amnestic and nonamnestic) groups.

CONCLUSION

Speech-based machine learning is powerful in detecting cognitive decline and probable AD dementia across a range of different feature domains, though important differences exist between these domains as well.

SUPPLEMENTAL MATERIAL

https://doi.org/10.23641/asha.23699733.

Embracing the uncertainty in human-machine collaboration to support clinical decision-making for mental health conditions

Two significant obstacles exist preventing the widespread usage of Deep Learning (DL) models for predicting healthcare outcomes in general and mental health conditions in particular. Firstly, DL models do not quantify the uncertainty in their predictions, so clinicians are unsure of which predictions they can trust. Secondly, DL models do not triage, i.e., separate which cases could be best handled by the human or the model. This paper attempts to address these obstacles using Bayesian Deep Learning (BDL), which extends DL probabilistically and allows us to quantify the model's uncertainty, which we use to improve human-model collaboration. We implement a range of state-of-the-art DL models for Natural Language Processing and apply a range of BDL methods to these models. Taking a step closer to the real-life scenarios of human-AI collaboration, we propose a Referral Learning methodology for the models that make predictions for certain instances while referring the rest of the instances to a human expert for further assessment. The study demonstrates that models can significantly enhance their performance by seeking human assistance in cases where the model exhibits high uncertainty, which is closely linked to misclassifications. Referral Learning offers two options: (1) supporting humans in cases where the model predicts with certainty, and (2) triaging cases where the model evaluated when it had a better chance of being right than the human by evaluating human disagreement. The latter method combines model uncertainty from BDL and human disagreement from multiple annotations, resulting in improved triaging capabilities.

ADscreen: A speech processing-based screening system for automatic identification of patients with Alzheimer's disease and related dementia

Alzheimer's disease and related dementias (ADRD) present a looming public health crisis, affecting roughly 5 million people and 11 % of older adults in the United States. Despite nationwide efforts for timely diagnosis of patients with ADRD, >50 % of them are not diagnosed and unaware of their disease. To address this challenge, we developed ADscreen, an innovative speech-processing based ADRD screening algorithm for the protective identification of patients with ADRD. ADscreen consists of five major components: (i) noise reduction for reducing background noises from the audio-recorded patient speech, (ii) modeling the patient's ability in phonetic motor planning using acoustic parameters of the patient's voice, (iii) modeling the patient's ability in semantic and syntactic levels of language organization using linguistic parameters of the patient speech, (iv) extracting vocal and semantic psycholinguistic cues from the patient speech, and (v) building and evaluating the screening algorithm. To identify important speech parameters (features) associated with ADRD, we used the Joint Mutual Information Maximization (JMIM), an effective feature selection method for high dimensional, small sample size datasets. Modeling the relationship between speech parameters and the outcome variable (presence/absence of ADRD) was conducted using three different machine learning (ML) architectures with the capability of joining informative acoustic and linguistic with contextual word embedding vectors obtained from the DistilBERT (Bidirectional Encoder Representations from Transformers). We evaluated the performance of the ADscreen on an audio-recorded patients' speech (verbal description) for the Cookie-Theft picture description task, which is publicly available in the dementia databank. The joint fusion of acoustic and linguistic parameters with contextual word embedding vectors of DistilBERT achieved F1-score = 84.64 (standard deviation [std] = ±3.58) and AUC-ROC = 92.53 (std = ±3.34) for training dataset, and F1-score = 89.55 and AUC-ROC = 93.89 for the test dataset. In summary, ADscreen has a strong potential to be integrated with clinical workflow to address the need for an ADRD screening tool so that patients with cognitive impairment can receive appropriate and timely care.

Learning implicit sentiments in Alzheimer's disease recognition with contextual attention features

Background

Alzheimer's disease (AD) is difficult to diagnose on the basis of language because of the implicit emotion of transcripts, which is defined as a supervised fuzzy implicit emotion classification at the document level. Recent neural network-based approaches have not paid attention to the implicit sentiments entailed in AD transcripts.

Method

A two-level attention mechanism is proposed to detect deep semantic information toward words and sentences, which enables it to attend to more words and fewer sentences differentially when constructing document representation. Specifically, a document vector was built by progressively aggregating important words into sentence vectors and important sentences into document vectors.

Results

Experimental results showed that our method achieved the best accuracy of 91.6% on annotated public Pitt corpora, which validates its effectiveness in learning implicit sentiment representation for our model.

Conclusion

The proposed model can qualitatively select informative words and sentences using attention layers, and this method also provides good inspiration for AD diagnosis based on implicit sentiment transcripts.

Ethical issues when using digital biomarkers and artificial intelligence for the early detection of dementia

Dementia poses a growing challenge for health services but remains stigmatized and under-recognized. Digital technologies to aid the earlier detection of dementia are approaching market. These include traditional cognitive screening tools presented on mobile devices, smartphone native applications, passive data collection from wearable, in-home and in-car sensors, as well as machine learning techniques applied to clinic and imaging data. It has been suggested that earlier detection and diagnosis may help patients plan for their future, achieve a better quality of life, and access clinical trials and possible future disease modifying treatments. In this review, we explore whether digital tools for the early detection of dementia can or should be deployed, by assessing them against the principles of ethical screening programs. We conclude that while the importance of dementia as a health problem is unquestionable, significant challenges remain. There is no available treatment which improves the prognosis of diagnosed disease. Progression from early-stage disease to dementia is neither given nor currently predictable. Available technologies are generally not both minimally invasive and highly accurate. Digital deployment risks exacerbating health inequalities due to biased training data and inequity in digital access. Finally, the acceptability of early dementia detection is not established, and resources would be needed to ensure follow-up and support for those flagged by any new system. We conclude that early dementia detection deployed at scale via digital technologies does not meet standards for a screening program and we offer recommendations for moving toward an ethical mode of implementation. This article is categorized under:Application Areas > Health CareCommercial, Legal, and Ethical Issues > Ethical ConsiderationsTechnologies > Artificial Intelligence.

Performance of machine learning algorithms for dementia assessment: impacts of language tasks, recording media, and modalities

OBJECTIVES

Automatic speech and language assessment methods (SLAMs) can help clinicians assess speech and language impairments associated with dementia in older adults. The basis of any automatic SLAMs is a machine learning (ML) classifier that is trained on participants' speech and language. However, language tasks, recording media, and modalities impact the performance of ML classifiers. Thus, this research has focused on evaluating the effects of the above-mentioned factors on the performance of ML classifiers that can be used for dementia assessment.

METHODOLOGY

Our methodology includes the following steps: (1) Collecting speech and language datasets from patients and healthy controls; (2) Using feature engineering methods which include feature extraction methods to extract linguistic and acoustic features and feature selection methods to select most informative features; (3) Training different ML classifiers; and (4) Evaluating the performance of ML classifiers to investigate the impacts of language tasks, recording media, and modalities on dementia assessment.

RESULTS

Our results show that (1) the ML classifiers trained with the picture description language task perform better than the classifiers trained with the story recall language task; (2) the data obtained from phone-based recordings improves the performance of ML classifiers compared to data obtained from web-based recordings; and (3) the ML classifiers trained with acoustic features perform better than the classifiers trained with linguistic features.

CONCLUSION

This research demonstrates that we can improve the performance of automatic SLAMs as dementia assessment methods if we: (1) Use the picture description task to obtain participants' speech; (2) Collect participants' voices via phone-based recordings; and (3) Train ML classifiers using only acoustic features. Our proposed methodology will help future researchers to investigate the impacts of different factors on the performance of ML classifiers for assessing dementia.

Deep learning for topical trend discovery in online discourse about Pre-Exposure Prophylaxis (PrEP)

Pre-Exposure Prophylaxis (PrEP) interventions are increasingly prevalent on social media. These data can be mined for insights about PrEP that may not be as apparent in surveys including personal musings about PrEP and barriers/facilitators to PrEP uptake. This study explores online discourse about PrEP using an interdisciplinary public health and computational informatics approach. We collected (N = 4,020) tweets using Twitter's Application Programming Interface (API). These data underwent a three-step neural network/deep learning process to identify clusters within these tweets and relative similarity/dissimilarity between clusters. We identified 25 distinct clusters from our original collection of tweets. These clusters represent general information about PrEP, how PrEP is communicated among diverse groups, and potential pockets of misinformation and disinformation regarding PrEP. Specific clusters of interest include discussions of medication side effects, social perception of PrEP usage, and concerns with costs and barriers to access of PrEP interventions. Our approach revealed diverse ways PrEP is contextualized online. Importantly this information can be leveraged to identify points of possible intervention for disinformation and misinformation about PrEP.

Use of Machine Learning in Stroke Rehabilitation: A Narrative Review

A narrative review was conducted of machine learning applications and research in the field of stroke rehabilitation. The machine learning models commonly used in medical research include random forest, logistic regression, and deep neural networks. Convolutional neural networks (CNNs), a type of deep neural network, are typically used for image analysis. Machine learning has been used in stroke rehabilitation to predict recovery of motor function using a large amount of clinical data as input. Recent studies on predicting motor function have trained CNN models using magnetic resonance images as input data together with clinical data to increase the accuracy of motor function prediction models. Additionally, a model interpreting videofluoroscopic swallowing studies was developed and investigated. In the future, we anticipate that machine learning will be actively used to treat stroke patients, such as predicting the occurrence of depression and the recovery of language, cognitive, and sensory function, as well as prescribing appropriate rehabilitation treatments.

A Computational Neural Network Model for College English Grammar Correction

For the error correction of English grammar, if there are errors in the semantic units (words and sentences), it will inevitably affect the subsequent text analysis and semantic understanding, and ultimately reduce the overall performance of the practical application system. Therefore, intelligent error detection and correction of the word and grammatical errors in English texts is one of the key and difficult points of natural language processing. This exploration innovatively combines a computational neural model with college grammar error correction to improve the accuracy of college grammar error correction. It studies the computational neural model in English grammar error correction based on a neural network named Knowledge and Neural machine translation powered College English Grammar Typo Correction (KNGTC). First, the Recurrent Neural Network is introduced, and the overall structure of the English grammatical error correction neural model is constructed. Moreover, the supervised training of Attention is discussed, and the experimental environment and experimental data are given. The results show that KNGTC has high accuracy in college English grammar correction, and the accuracy of this model in CET-4 and CET-6 writing can reach 82.69%. The English grammar error correction model based on the computational neural network has perfect function and strong error correction ability. The optimization and perfection of the model can improve students' English grammar level, which has certain practical value. After years of continuous optimization and improvement, English grammar error correction technology has entered a performance bottleneck. This mode's construction can break the current technology's limitations and bring a better user experience. Therefore, it is very valuable to study the error correction model of English grammar in practical application.

Unboxing Deep Learning Model of Food Delivery Service Reviews Using Explainable Artificial Intelligence (XAI) Technique

The demand for food delivery services (FDSs) during the COVID-19 crisis has been fuelled by consumers who prefer to order meals online and have it delivered to their door than to wait at a restaurant. Since many restaurants moved online and joined FDSs such as Uber Eats, Menulog, and Deliveroo, customer reviews on internet platforms have become a valuable source of information about a company’s performance. FDS organisations strive to collect customer complaints and effectively utilise the information to identify improvements needed to enhance customer satisfaction. However, only a few customer opinions are addressed because of the large amount of customer feedback data and lack of customer service consultants. Organisations can use artificial intelligence (AI) instead of relying on customer service experts and find solutions on their own to save money as opposed to reading each review. Based on the literature, deep learning (DL) methods have shown remarkable results in obtaining better accuracy when working with large datasets in other domains, but lack explainability in their model. Rapid research on explainable AI (XAI) to explain predictions made by opaque models looks promising but remains to be explored in the FDS domain. This study conducted a sentiment analysis by comparing simple and hybrid DL techniques (LSTM, Bi-LSTM, Bi-GRU-LSTM-CNN) in the FDS domain and explained the predictions using SHapley Additive exPlanations (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME). The DL models were trained and tested on the customer review dataset extracted from the ProductReview website. Results showed that the LSTM, Bi-LSTM and Bi-GRU-LSTM-CNN models achieved an accuracy of 96.07%, 95.85% and 96.33%, respectively. The model should exhibit fewer false negatives because FDS organisations aim to identify and address each and every customer complaint. The LSTM model was chosen over the other two DL models, Bi-LSTM and Bi-GRU-LSTM-CNN, due to its lower rate of false negatives. XAI techniques, such as SHAP and LIME, revealed the feature contribution of the words used towards positive and negative sentiments, which were used to validate the model.

Deep Learning-Based Diagnosis of Alzheimer’s Disease

Alzheimer’s disease (AD), the most familiar type of dementia, is a severe concern in modern healthcare. Around 5.5 million people aged 65 and above have AD, and it is the sixth leading cause of mortality in the US. AD is an irreversible, degenerative brain disorder characterized by a loss of cognitive function and has no proven cure. Deep learning techniques have gained popularity in recent years, particularly in the domains of natural language processing and computer vision. Since 2014, these techniques have begun to achieve substantial consideration in AD diagnosis research, and the number of papers published in this arena is rising drastically. Deep learning techniques have been reported to be more accurate for AD diagnosis in comparison to conventional machine learning models. Motivated to explore the potential of deep learning in AD diagnosis, this study reviews the current state-of-the-art in AD diagnosis using deep learning. We summarize the most recent trends and findings using a thorough literature review. The study also explores the different biomarkers and datasets for AD diagnosis. Even though deep learning has shown promise in AD diagnosis, there are still several challenges that need to be addressed.

A Transfer Learning Method for Detecting Alzheimer's Disease Based on Speech and Natural Language Processing

Alzheimer's disease (AD) is a neurodegenerative disease that is difficult to be detected using convenient and reliable methods. The language change in patients with AD is an important signal of their cognitive status, which potentially helps in early diagnosis. In this study, we developed a transfer learning model based on speech and natural language processing (NLP) technology for the early diagnosis of AD. The lack of large datasets limits the use of complex neural network models without feature engineering, while transfer learning can effectively solve this problem. The transfer learning model is firstly pre-trained on large text datasets to get the pre-trained language model, and then, based on such a model, an AD classification model is performed on small training sets. Concretely, a distilled bidirectional encoder representation (distilBert) embedding, combined with a logistic regression classifier, is used to distinguish AD from normal controls. The model experiment was evaluated on Alzheimer's dementia recognition through spontaneous speech datasets in 2020, including the balanced 78 healthy controls (HC) and 78 patients with AD. The accuracy of the proposed model is 0.88, which is almost equivalent to the champion score in the challenge and a considerable improvement over the baseline of 75% established by organizers of the challenge. As a result, the transfer learning method in this study improves AD prediction, which does not only reduces the need for feature engineering but also addresses the lack of sufficiently large datasets.

Improving Alzheimer's Disease Detection for Speech Based on Feature Purification Network

Alzheimer's disease (AD) is a neurodegenerative disease involving the decline of cognitive ability with illness progresses. At present, the diagnosis of AD mainly depends on the interviews between patients and doctors, which is slow, expensive, and subjective, so it is not a better solution to recognize AD using the currently available neuropsychological examinations and clinical diagnostic criteria. A recent study has indicated the potential of language analysis for AD diagnosis. In this study, we proposed a novel feature purification network that can improve the representation learning of transformer model further. Though transformer has made great progress in generating discriminative features because of its long-distance reasoning ability, there is still room for improvement. There exist many common features that are not indicative of any specific class, and we rule out the influence of common features from traditional features extracted by transformer encoder and can get more discriminative features for classification. We apply this method to improve transformer's performance on three public dementia datasets and get improved classification results markedly. Specifically, the method on Pitt datasets gets state-of-the-art (SOTA) result.

Transfer learning for non-image data in clinical research: A scoping review

BACKGROUND

Transfer learning is a form of machine learning where a pre-trained model trained on a specific task is reused as a starting point and tailored to another task in a different dataset. While transfer learning has garnered considerable attention in medical image analysis, its use for clinical non-image data is not well studied. Therefore, the objective of this scoping review was to explore the use of transfer learning for non-image data in the clinical literature.

METHODS AND FINDINGS

We systematically searched medical databases (PubMed, EMBASE, CINAHL) for peer-reviewed clinical studies that used transfer learning on human non-image data. We included 83 studies in the review. More than half of the studies (63%) were published within 12 months of the search. Transfer learning was most often applied to time series data (61%), followed by tabular data (18%), audio (12%) and text (8%). Thirty-three (40%) studies applied an image-based model to non-image data after transforming data into images (e.g. spectrograms). Twenty-nine (35%) studies did not have any authors with a health-related affiliation. Many studies used publicly available datasets (66%) and models (49%), but fewer shared their code (27%).

CONCLUSIONS

In this scoping review, we have described current trends in the use of transfer learning for non-image data in the clinical literature. We found that the use of transfer learning has grown rapidly within the last few years. We have identified studies and demonstrated the potential of transfer learning in clinical research in a wide range of medical specialties. More interdisciplinary collaborations and the wider adaption of reproducible research principles are needed to increase the impact of transfer learning in clinical research.