ABSTRACT
Background: The growing number of older adults in recent decades has led to more prevalent geriatric diseases, such as strokes and dementia. Therefore, Alzheimer's disease (AD), as the most common type of dementia, has become more frequent too. Background: Objective: The goals of this work are to present state-of-the-art studies focused on the automatic diagnosis and prognosis of AD and its early stages, mainly mild cognitive impairment, and predicting how the research on this topic may change in the future. Methods: Articles found in the existing literature needed to fulfill several selection criteria. Among others, their classification methods were based on artificial neural networks (ANNs), including deep learning, and data not from brain signals or neuroimaging techniques were used. Considering our selection criteria, 42 articles published in the last decade were finally selected. Results: The most medically significant results are shown. Similar quantities of articles based on shallow and deep ANNs were found. Recurrent neural networks and transformers were common with speech or in longitudinal studies. Convolutional neural networks (CNNs) were popular with gait or combined with others in modular approaches. Above one third of the cross-sectional studies utilized multimodal data. Non-public datasets were frequently used in cross-sectional studies, whereas the opposite in longitudinal ones. The most popular databases were indicated, which will be helpful for future researchers in this field. Conclusions: The introduction of CNNs in the last decade and their superb results with neuroimaging data did not negatively affect the usage of other modalities. In fact, new ones emerged.
Subject(s)
Alzheimer Disease , Early Diagnosis , Neural Networks, Computer , Neuroimaging , Humans , Alzheimer Disease/diagnostic imaging , Alzheimer Disease/diagnosis , Prognosis , Neuroimaging/methods , Biomarkers , Cognitive Dysfunction/diagnostic imaging , Cognitive Dysfunction/diagnosis , Deep LearningABSTRACT
Clinical procedure for mild cognitive impairment (MCI) is mainly based on clinical records and short cognitive tests. However, low suspicion and difficulties in understanding test cut-offs make diagnostic accuracy being low, particularly in primary care. Artificial neural networks (ANNs) are suitable to design computed aided diagnostic systems because of their features of generating relationships between variables and their learning capability. The main aim pursued in that work is to explore the ability of a hybrid ANN-based system in order to provide a tool to assist in the clinical decision-making that facilitates a reliable MCI estimate. The model is designed to work with variables usually available in primary care, including Minimental Status Examination (MMSE), Functional Assessment Questionnaire (FAQ), Geriatric Depression Scale (GDS), age, and years of education. It will be useful in any clinical setting. Other important goal of our study is to compare the diagnostic rendering of ANN-based system and clinical physicians. A sample of 128 MCI subjects and 203 controls was selected from the Alzheimer's Disease Neuroimaging Initiative (ADNI). The ANN-based system found the optimal variable combination, being AUC, sensitivity, specificity, and clinical utility index (CUI) calculated. The ANN results were compared with those from medical experts which include two family physicians, a neurologist, and a geriatrician. The optimal ANN model reached an AUC of 95.2%, with a sensitivity of 90.0% and a specificity of 84.78% and was based on MMSE, FAQ, and age inputs. As a whole, physician performance achieved a sensitivity of 46.66% and a specificity of 91.3%. CUIs were also better for the ANN model. The proposed ANN system reaches excellent diagnostic accuracy although it is based only on common clinical tests. These results suggest that the system is especially suitable for primary care implementation, aiding physicians work with cognitive impairment suspicions.
Subject(s)
Cognitive Dysfunction/diagnosis , Decision Support Systems, Clinical , Diagnosis, Computer-Assisted/methods , Neural Networks, Computer , Neuropsychological Tests , Aged , Aged, 80 and over , Area Under Curve , Case-Control Studies , Cognitive Dysfunction/psychology , Computational Biology , Databases, Factual/statistics & numerical data , Decision Support Systems, Clinical/statistics & numerical data , Diagnosis, Computer-Assisted/statistics & numerical data , Humans , Neuropsychological Tests/statistics & numerical data , Sensitivity and SpecificityABSTRACT
Benzimidazole fungicides (BFs) are a type of pesticide of high environmental interest characterized by a heavy fluorescence spectral overlap which complicates its detection in mixtures. In this paper, we present a computational study based on supervised neural networks for a multi-label classification problem. Specifically, backpropagation networks (BPNs) with data fusion and ensemble schemes are used for the simultaneous resolution of difficult multi-fungicide mixtures. We designed, optimized and compared simple BPNs, BPNs with data fusion and BPNs ensembles. The information environment used is made up of synchronous and conventional BF fluorescence spectra. The mixture spectra are not used in the training nor the validation stage. This study allows us to determine the convenience of fusioning the labels of carbendazim and benomyl for the identification of BFs in complex multi-fungicide mixtures.
Subject(s)
Benzimidazoles/chemistry , Fungicides, Industrial/analysis , Fluorescence , Neural Networks, ComputerABSTRACT
In this paper, we approach, using neural computation and ensemble systems, a pattern classification problem in fluorescence spectrometry, the resolution of difficult multi-fungicide mixtures (overlapping), specifically the benzimidazole fungicides, benomyl, carbendazim, thiabendazole and fuberidazole. These fungicides are compounds of an important environmental interest. Because of this, from an analytical point of view, it is interesting to develop sensitive, selective and simple methods for their determination. Fluorescence spectrometry has proven to be a sensitive and selective technique for determination of many compounds of environmental interest, but in some cases it is not enough. HUMANN is a hierarchical, unsupervised, modular, adaptive neural net with high biological plausibility, which has shown to be suitable for identification of these fungicides and organochlorinated compounds of environmental interest. We propose two modular artificial intelligent systems, with a structure of pre-processing and processing stage, a multi-input HUMANN-based system, using multi-fluorescence spectra as input to the system, and a HUMANN-ensemble system. We analyze the optimal configuration of inputs and the ensemble in order to obtain better results. We study such figures as precision and sensitivity of the method. Our proposal is a smart, flexible and effective complementary method, which allows reducing the analytical and/or computational complexity of the analysis.