Terms Of Prediction Research Articles

Spatial-Intensity Transforms for Medical Image-to-Image Translation.

Image-to-image translation has seen major advances in computer vision but can be difficult to apply to medical images, where imaging artifacts and data scarcity degrade the performance of conditional generative adversarial networks. We develop the spatial-intensity transform (SIT) to improve output image quality while closely matching the target domain. SIT constrains the generator to a smooth spatial transform (diffeomorphism) composed with sparse intensity changes. SIT is a lightweight, modular network component that is effective on various architectures and training schemes. Relative to unconstrained baselines, this technique significantly improves image fidelity, and our models generalize robustly to different scanners. Additionally, SIT provides a disentangled view of anatomical and textural changes for each translation, making it easier to interpret the model's predictions in terms of physiological phenomena. We demonstrate SIT on two tasks: predicting longitudinal brain MRIs in patients with various stages of neurodegeneration, and visualizing changes with age and stroke severity in clinical brain scans of stroke patients. On the first task, our model accurately forecasts brain aging trajectories without supervised training on paired scans. On the second task, it captures associations between ventricle expansion and aging, as well as between white matter hyperintensities and stroke severity. As conditional generative models become increasingly versatile tools for visualization and forecasting, our approach demonstrates a simple and powerful technique for improving robustness, which is critical for translation to clinical settings. Source code is available at github.com/clintonjwang/spatial-intensity-transforms.

Multi-prediction of electric load and photovoltaic solar power in grid-connected photovoltaic system using state transition method

In the grid-connected photovoltaic system (GPVS), due to characteristics of fluctuation and intermittency for photovoltaic solar power, and high randomness for electric load, it is of great difficulty for integrating photovoltaic solar power into power grid. Therefore, an accurate prediction of short-term electric load and photovoltaic solar power is of great importance for balancing supply and demand. Currently, numerous isolated models about the forecasting of electric load and photovoltaic solar power have emerged, while the coupling effect between them has been hardly considered and lower stability of existing methods brings great difficulty in providing reliable predictions at practical applications. To address this gap, this paper proposes an interpretable multi-prediction model for short-term (day-ahead) electric load and photovoltaic solar power forecasting. In the framework, a non-parametric functional principal component analysis (FPCA) is constructed to extract the overall trend and identify dominant modes of variation in the daily electric load and photovoltaic solar power data. Furthermore, state transition matrix is proposed to comprehensively interpret the coupling effect, with which a novel multi-prediction strategy that takes advantage of coupling effect is further introduced, where the Maximum Likelihood Estimation (MLE) is employed to estimate unknown parameters. Moreover, data from California Independent System Operator (ISO) is utilized to investigate the performance of proposed method, and its results are compared with those from other widely-used techniques. Results show that the proposed method can increase prediction accuracy of electric load and photovoltaic solar power by 16.84% and 10.57%, respectively, with narrow fluctuations and reasonable computational cost (211.11 s), demonstrating that it can provide better predictions in terms of prediction accuracy, stability and applicability.

Evaluating the accuracy of predicted bridge condition using machine learning: the role of condition history

Effective maintenance decisions for bridges depend on accurate performance prediction. Machine learning (ML) models use historical bridge performance data to learn and predict performance. However, in many agencies, the condition history of bridges is limited and does not go beyond a few years. The question, therefore, is, to what extent does condition history help us make better predictions? To address this question, a ML model was developed that analysed more than 600,000 bridge decks with 27 years of condition history. Two data selection methods were designed: non-overlapping and overlapping data. The non-overlapping data are typically used to train the model. The overlapping data introduced in this study uses the data more efficiently for model training recognising that strings of historical data convey more information. Longer term predictions were found to be positively impacted by every additional year of condition history. Short-term condition prediction (one or two years) does not need significant historical data. It was also found that overlapping data, compared to non-overlapping data, produced larger training samples and had higher prediction accuracy in the majority of experiments, but at the cost of higher running time due to a larger sample size.

Mechanical Response of an Industrial Piping System Under Strong Cyclic Loading

Abstract The paper presents a combined experimental and numerical investigation of cyclic loading response of an internally pressurized 8-inch-diameter steel piping system. The piping system comprises three elbows and is subjected to quasi-static end-displacement excitation. Global deformation and local strain measurements are obtained, indicating significant strain ratcheting at critical locations of the elbows. The piping system failed under low-cycle fatigue undergoing through-thickness cracking at the flank of the most strained elbow, after 129 loading cycles with measured strain range of approximately 3%. Postfatigue metallographic examination of the elbows indicated that fatigue cracking initiates from the inner surface of the pipe elbow. In all elbows, several microcracks develop along the inner surface of elbow flanks, whereas the outer surface remained practically intact before through-thickness cracking occurred. Finite element simulations, with a cyclic-plasticity model calibrated properly in terms of small-scale material tests, provide very good predictions in terms of local strain evolution at critical locations. Numerical results at the intrados and the extrados of the critical elbow of the piping system verify strain ratcheting and the location of crack initiation observed in the experiments. This paper can be used as a reference for future experiments on cyclic loading of piping components, and for benchmarking constitutive modeling for simulating ratcheting.

An explainable deep learning approach for detection and isolation of sensor and machine faults in predictive maintenance paradigm

The predictive health maintenance techniques identify the machine faults by analyzing the data collected by low-cost sensors assuming that sensors are free from any faults. However, aging and environmental condition cause sensors also be faulty, leading to incorrect interpretations of the collected data and subsequently resulting in erroneous machine health predictions. To mitigate this problem, this paper proposes a hybrid model that can differentiate between sensor and system faults. The data used for training the model is collected from a power system hardware setup by experimental procedures. A convolutional neural network (CNN) model is used to extract optimized features from the raw data automatically, which are then fed to the eXtreme Gradient Boosting (XGBoost) model for sensor and machine fault isolation with an overall accuracy of 98.15%. The data having sensor fault was then fed to a deep autoencoder, which eliminated the sensor fault components and reconstructed the data with an average root mean square error of 0.0576. Thereafter, the corrected signal was used to detect the system fault using the hybrid CNN-XGBoost model with 99.77% accuracy. Therefore, by isolating the sensor faults, the proposed technique establishes better confidence in predictive maintenance. Further, explainable AI has been utilized to interpret the model prediction in human-understandable terms in order to increase trustworthiness.

AffinityVAE: A multi-objective model for protein-ligand affinity prediction and drug design

In the prediction of protein-ligand affinity, the traditional methods require a large amount of computing resources, and have certain limitations in predicting and simulating the structural changes. Although employing data-driven approaches can yield favorable outcomes in deep learning, it entails a lack of interpretability. Some methods may require additional structural information or domain knowledge to support the interpretation, which may limit their applicability. This paper proposes an affinity variational autoencoder (AffinityVAE) using interaction feature mapping and a variational autoencoder, which consists of a multi-objective model capable of end-to-end affinity prediction and drug discovery. In this study, the limitations of affinity prediction in terms of interpretability are tackled by proposing the concept of a protein-ligand interaction feature map. This increases the diversity and quantity of protein-ligand binding data by designing an adaptive autoencoder of target chemical properties to generate new ligands similar to known ligands and adding them to the original training set. AffinityVAE is then retrained using this extended training set to further validate the protein-ligand binding affinity prediction. Comparisons were conducted between the AffinityVAE and recent methods to demonstrate the high efficiency of the proposed model. The experimental results show that AffinityVAE has very high prediction performance, and it has the potential to enhance the diversity and the amount of protein-ligand binding data, which promotes the drug development.

Usage of Machine Learning Algorithms for Establishing an Effective Protocol for the In Vitro Micropropagation Ability of Black Chokeberry (Aronia melanocarpa (Michx.) Elliott)

The primary objective of this research was to ascertain the optimal circumstances for the successful growth of black chokeberry (Aronia melanocarpa (Michx.) Elliott) using tissue culture techniques. Additionally, the study aimed to explore the potential use of machine learning algorithms in this context. The present research investigated a range of in vitro parameters such as total number of roots (TNR), longest root length (LRL), average root length (ARL), number of main roots (NMR), number of siblings (NS), shoot length (SL), shoot diameter (SD), leaf width (LW), and leaf length (LL) for Aronia explants cultivated in different media (Murashige and Skoog (MS) and woody plant medium (WPM)) with different concentrations (0, 0.5, 1, 1.5, and 2 mg L−1) of indole-3-butyric acid (IBA). The study showed that IBA hormone levels may affect WPM properties, affecting the LRL and ARL variables. Aronia explant media treated with 2 mg L−1 IBA had the greatest TNR, NMR, NS, SL, and SD values; 31.67 pieces, 2.37 pieces, 5.25 pieces, 66.60 mm, and 2.59 mm, in that order. However, Aronia explants treated with 1 mg L−1 IBA had the highest LW (9.10 mm) and LL (14.58 mm) values. Finally, Aronia explants containing 0.5 mg L−1 IBA had the greatest LRL (89.10 mm) and ARL (57.57 mm) values. In general, the results observed (TNR, LRL, ARL, NMR, NS, SL, SD, LW, and LL) indicate that Aronia explants exhibit superior growth and development in WPM (25.68 pieces, 68.10 mm, 51.64 mm, 2.17 pieces, 4.33 pieces, 57.95 mm, 2.49 mm, 8.08 mm, and 14.26 mm, respectively) as opposed to MS medium (20.27 pieces, 59.92 mm, 47.25 mm, 1.83 pieces, 3.57 pieces, 49.34 mm, 2.13 mm, 6.99 mm, and 12.21 mm, respectively). In the context of the in vitro culturing of Aronia explants utilizing MS medium and WPM, an analysis of machine learning models revealed that the XGBoost and SVM models perform better than the RF, KNN, and GP models when it comes to making predictions about those variables. In particular, the XGBoost model stood out due to the fact that it had the greatest R-squared value, and showed higher predictive ability in terms of properly forecasting values in comparison to actual outcomes. The findings of a linear regression (LR) analysis were used in order to conduct an efficacy study of the XGBoost model. The LR results especially confirmed the findings for the SD, NS, and NMR variables, whose R-squared values were more than 0.7. This demonstrates the extraordinary accuracy that XGboost has in predicting these particular variables. As a consequence of this, it is anticipated that it will be beneficial to make use of the XGboost model in the dosage optimization and estimation of in vitro parameters in micropropagation studies of the Aronia plant for further scientific investigation.

Game-theoretical description of the go-or-grow dichotomy in tumor development for various settings and parameter constellations

A medically important feature of several types of tumors is their ability to “decide” between staying at a primary site in the body or leaving it and forming metastases. The present theoretical study aims to provide a better understanding of the ultimate reasons for this so-called “go-or-grow” dichotomy. To that end, we use game theory, which has proven to be useful in analyzing the competition between tumors and healthy tissues or among different tumor cells. We begin by determining the game types in the Basanta–Hatzikirou–Deutsch model, depending on the parameter values. Thereafter, we suggest and analyze five modified variants of the model. For example, in the basic model, the deadlock game, Prisoner’s Dilemma, and hawk-dove game can occur. The modified versions lead to several additional game types, such as battle of the sexes, route-choice, and stag-hunt games. For some game types, all cells are predicted to stay on their original site (“grow phenotype”), while for other types, only a certain fraction stay and the other cells migrate away (“go phenotype”). If the nutrient supply at a distant site is high, all the cells are predicted to go. We discuss our predictions in terms of the pros and cons of caloric restriction and limitations of the supply of vitamins or methionine. Our results may help devise treatments to prevent metastasis.

Nonlocal Theory for Submerged Cantilever Beams Undergoing Torsional Vibrations

Abstract We propose a new theory for fluid–structure interactions of cantilever microbeams undergoing small amplitude vibrations in viscous fluids. The method is based on the concept of nonlocal modal hydrodynamic functions that accurately capture three-dimensional (3D) fluid loading on the structure. For short beams for which 3D effects become prominent, existing local theories based on two-dimensional (2D) fluid approximations are inadequate to predict the dynamic response. We discuss and compare model predictions in terms of frequency response functions, modal shapes, quality factors, and added mass ratios with the predictions of the local theory, and we validate our new model with experimental results.

Deep Learning to Predict the Cell Proliferation and Prognosis of Non-Small Cell Lung Cancer Based on FDG-PET/CT Images.

(1) Background: Cell proliferation (Ki-67) has important clinical value in the treatment and prognosis of non-small cell lung cancer (NSCLC). However, current detection methods for Ki-67 are invasive and can lead to incorrect results. This study aimed to explore a deep learning classification model for the prediction of Ki-67 and the prognosis of NSCLC based on FDG-PET/CT images. (2) Methods: The FDG-PET/CT scan results of 159 patients with NSCLC confirmed via pathology were analyzed retrospectively, and the prediction models for the Ki-67 expression level based on PET images, CT images and PET/CT combined images were constructed using Densenet201. Based on a Ki-67 high expression score (HES) obtained from the prediction model, the survival rate of patients with NSCLC was analyzed using Kaplan-Meier and univariate Cox regression. (3) Results: The statistical analysis showed that Ki-67 expression was significantly correlated with clinical features of NSCLC, including age, gender, differentiation state and histopathological type. After a comparison of the three models (i.e., the PET model, the CT model, and the FDG-PET/CT combined model), the combined model was found to have the greatest advantage in Ki-67 prediction in terms of AUC (0.891), accuracy (0.822), precision (0.776) and specificity (0.902). Meanwhile, our results indicated that HES was a risk factor for prognosis and could be used for the survival prediction of NSCLC patients. (4) Conclusions: The deep-learning-based FDG-PET/CT radiomics classifier provided a novel non-invasive strategy with which to evaluate the malignancy and prognosis of NSCLC.

Uncovering the System Vulnerability and Criticality of Human Brain Under Dynamical Neuropathological Events in Alzheimer's Disease.

Despite the striking efforts in investigating neurobiological factors behind the acquisition of amyloid-β (A), protein tau (T), and neurodegeneration ([N]) biomarkers, the mechanistic pathways of how AT[N] biomarkers spreading throughout the brain remain elusive. To disentangle the massive heterogeneities in Alzheimer's disease (AD) progressions and identify vulnerable/critical brain regions to AD pathology. In this work, we characterized the interaction of AT[N] biomarkers and their propagation across brain networks using a novel bistable reaction-diffusion model, which allows us to establish a new systems biology underpinning of AD progression. We applied our model to large-scale longitudinal neuroimages from the ADNI database and studied the systematic vulnerability and criticality of brains. Our model yields long term prediction that is statistically significant linear correlated with temporal imaging data, produces clinically consistent risk prediction, and captures the Braak-like spreading pattern of AT[N] biomarkers in AD development. Our major findings include (i) tau is a stronger indicator of regional risk compared to amyloid, (ii) temporal lobe exhibits higher vulnerability to AD-related pathologies, (iii) proposed critical brain regions outperform hub nodes in transmitting disease factors across the brain, and (iv) comparing the spread of neuropathological burdens caused by amyloid-β and tau diffusions, disruption of metabolic balance is the most determinant factor contributing to the initiation and progression of AD.

Weight prediction and recognition of latent subject terms based on the fusion of explicit & implicit information about keyword

Weight prediction and recognition of subject term in documents is widely used in application fields such as literature recommendation and keyword retrieval. Unlike traditional methods which only focus on searching for subject terms from existing keywords in documents, this paper proposes a recognition method of latent subject term outside documents based on a one-class collaborative filtering algorithm that fuses explicit and implicit information. A matrix factorization model based on the analysis of document activity and subject term popularity is constructed to measure the correlation probability between documents and subject terms that do not appear in the current document. After these subject terms being divided into Latent Subject Terms (LST) and Irrelevant Subject Terms (IST), two methods for predicting the weight of these types of subject terms is introduced, which include Hybrid Filling with Preference Coefficients (HFPC), and Zero Filling. In order to verify the effectiveness of subject term recognition, many collaborative filtering recommendation algorithms are conducted with the filled document keyword matrix. On the basis of not changing these algorithms, MAE and FCP can be improved by 28.01% and 22.79% at most, while P@N and NDCG@N can be improved by 22.37% and 27.06%, respectively.

"What" and "when" predictions modulate auditory processing in a mutually congruent manner.

Extracting regularities from ongoing stimulus streams to form predictions is crucial for adaptive behavior. Such regularities exist in terms of the content of the stimuli and their timing, both of which are known to interactively modulate sensory processing. In real-world stimulus streams such as music, regularities can occur at multiple levels, both in terms of contents (e.g., predictions relating to individual notes vs. their more complex groups) and timing (e.g., pertaining to timing between intervals vs. the overall beat of a musical phrase). However, it is unknown whether the brain integrates predictions in a manner that is mutually congruent (e.g., if "beat" timing predictions selectively interact with "what" predictions falling on pulses which define the beat), and whether integrating predictions in different timing conditions relies on dissociable neural correlates. To address these questions, our study manipulated "what" and "when" predictions at different levels - (local) interval-defining and (global) beat-defining - within the same stimulus stream, while neural activity was recorded using electroencephalogram (EEG) in participants (N = 20) performing a repetition detection task. Our results reveal that temporal predictions based on beat or interval timing modulated mismatch responses to violations of "what" predictions happening at the predicted time points, and that these modulations were shared between types of temporal predictions in terms of the spatiotemporal distribution of EEG signals. Effective connectivity analysis using dynamic causal modeling showed that the integration of "what" and "when" predictions selectively increased connectivity at relatively late cortical processing stages, between the superior temporal gyrus and the fronto-parietal network. Taken together, these results suggest that the brain integrates different predictions with a high degree of mutual congruence, but in a shared and distributed cortical network. This finding contrasts with recent studies indicating separable mechanisms for beat-based and memory-based predictive processing.

KIT reactor safety research for LWRs: Research lines, numerical tools, and prospects

This paper describes the main research lines of the Karlsruhe Institute of Technology (KIT) relevant to Light Water Reactors (LWR) that permit both the safety evaluation of different reactor designs and the prediction of the radiological source term including its impact in the case of hypothetical severe accidents. The numerical tools for core analysis, nuclear power plant analysis of design basis and severe accidents, and the estimation of the radioactive dispersions of fission products after a core meltdown accident are described and selected results are discussed. The analytical investigations at KIT are complemented by experimental investigations at two facilities, namely the COSMOS-H/-L and the QUENCH facilities devoted to safety-relevant Thermal-Hydraulic (TH) phenomena and in-vessel severe accident early-phase for different kinds of fuel rods including Accident Tolerant Fuel (ATF). These data are crucial for the validation of the numerical tools used for safety demonstration. The paper is complemented by a brief description of the in-house tools under development for improved core/plant analysis based on the multi-physics/-scale coupling methodologies, where the in-house tools are combined with external system TH, Monte Carlo, and thermo-mechanic code to increase the prediction accuracy of the core/plant behaviour under stationary and accidental conditions. Selected results are presented and discussed. It can be stated that the long-lasting safety research activities at KIT devoted to LWR led to the development of a computation route combining external with in-house codes that is very much appropriate for safety evaluations and risk assessment; it includes unique multi-physics/-scale coupled tools for improved core and plant analysis of different reactor designs, e.g. PWR, VVER, BWR, SMR, and research reactors.

Investigation of the NH3 consumption and H2O/NO formation during the oxidation of n-heptane and iso-octane blended with ammonia

Adding highly active fuels with ammonia is a compromising way to increase the mixture's reactivity. However, there are very few studies blending ammonia with large hydrocarbons, such as n-heptane and isooctane. In this study, the NH3, NO, and H2O time-histories were simultaneously measured during the oxidation of ammonia/n-heptane and ammonia/isooctane mixtures (equivalence ratio ϕ ∼ 1.2) in a shock tube. For the ammonia/n-heptane mixture, a reduced kinetics model, the Wang 2023 model, shows the best predictions in terms of the mixture reactivity and the H2O production but underpredicts the NO plateau. For the ammonia/isooctane mixture, the CRECK_TPRF_Alcohols_NOx (CRECK) model is the only available kinetics model containing ammonia and isooctane mechanisms. The CRECK model captures the H2O plateau but underpredicts the ammonia/isooctane mixture reactivity and the NO plateau. Detailed analyses are conducted based on the Wang 2023 and CRECK models. Rate-of-production (ROP) analyses reveal that: the consumption or formation pathways of the three species in the two models are basically the same but the magnitudes of the ROPs are quite different, especially for the NO formation; the reaction of ethylene or propylene with NH2 radical in the Wang 2023 model also contributes significantly to the NH3 consumption but they are not considered in the CRECK model. Sensitivity analyses show that: H + O2 = O + OH is the most promoting reaction for both mixtures and models; however, several reactions show strong effects in one mixture while having little effect in another mixture, or even playing an opposite effect. After modifying or adding some key reactions, the Wang 2023 and CRECK models get significantly improved in predicting the species' time-history during the oxidation of the ammonia/n-heptane or ammonia/isooctane mixture.

Predicting Age of Acquisition for Children's Early Vocabulary in Five Languages Using Language Model Surprisal.

What makes a word easy to learn? Early-learned words are frequent and tend to name concrete referents. But words typically do not occur in isolation. Some words are predictable from their contexts; others are less so. Here, we investigate whether predictability relates to when children start producing different words (age of acquisition; AoA). We operationalized predictability in terms of a word's surprisal in child-directed speech, computed using n-gram and long-short-term-memory (LSTM) language models. Predictability derived from LSTMs was generally a better predictor than predictability derived from n-gram models. Across five languages, average surprisal was positively correlated with the AoA of predicates and function words but not nouns. Controlling for concreteness and word frequency, more predictable predicates and function words were learned earlier. Differences in predictability between languages were associated with cross-linguistic differences in AoA: the same word (when it was a predicate) was produced earlier in languages where the word was morepredictable.

Distributional prediction of short-term traffic using neural networks

Neural network (NN)-based models have recently achieved outstanding results in short-term traffic prediction. However, most of these are based on the regression approach and trained to generate a single data point as a predicted value for future timesteps, which does not provide information on prediction uncertainty and limits its performance under different traffic conditions. To solve this problem, this study proposes a novel, high-dimensional distributional prediction (HDP) framework. This method has been validated by a series of experiments using the Caltrans Performance Measurement System dataset and four widely used NN models. The results suggest that the proposed HDP scheme can help existing NN structures to (1) generate adaptive distributional predictions for quantifying the uncertainty of multiple targets, and (2) gain better point prediction in terms of accuracy and robustness. Furthermore, we demonstrate that predicted speed distributions can be used for travel time estimation, outperforming other traditional methods in unexpected traffic conditions such as traffic incidents.

Financial uncertainty and stock market volatility

AbstractThis study explores the relation between financial uncertainty and volatility in China. The time variation in financial uncertainty shocks is theoretically closely related to stock return dynamics. Empirically, the financial uncertainty measure is based on a large set of economic and financial variables and captures its unpredictable component. Over the sample period from 2000 to 2021, we find that financial uncertainty positively impacts the trend component of market volatility and that it improves volatility predictions in both statistical and economic terms. Our study sheds new light on the sources driving volatility and the dynamic relation between uncertainty and volatility components.

Interpretation and Prediction of the CO2 Sequestration of Steel Slag by Machine Learning.

The utilization of steel slag for CO2 sequestration is an effective way to reduce carbon emissions. The reactivity of steel slag in CO2 sequestration depends mainly on material and process parameters. However, there are many puzzles in regard to practical applications due to the different evaluations of process parameters and the lack of investigation of material parameters. In this study, 318 samples were collected to investigate the interactive influence of 12 factors on the carbonation reactivity of steel slag by machine learning with SHapley Additive exPlanations (SHAP). Multilayer perceptron (MLP), random forest, and support vector regression models were built to predict the slurry-phase CO2 sequestration of steel slag. The MLP model performed well in terms of prediction ability and generalization with comprehensive interpretability. The SHAP results showed that the impact of the process parameters was greater than that of the material parameters. Interestingly, the iron ore phase of steel slag was revealed to have a positive effect on steel slag carbonation by SHAP analysis. Combined with previous literature, the carbonation mechanism of steel slag was proposed. Quantitative analysis based on SHAP indicated that steel slag had good carbonation reactivity when the mass fractions of "CaO + MgO", "SiO2 + Al2O3", "Fe2O3", and "MnO" varied from 50-55%, 10-15%, 30-35%, and <5%, respectively.

Improving algal bloom detection using spectroscopic analysis and machine learning: A case study in a large artificial reservoir, South Korea

The prediction of algal blooms using traditional water quality indicators is expensive, labor-intensive, and time-consuming, making it challenging to meet the critical requirement of timely monitoring for prompt management. Using optical measures for forecasting algal blooms is a feasible and useful method to overcome these problems. This study explores the potential application of optical measures to enhance algal bloom prediction in terms of prediction accuracy and workload reduction, aided by machine learning (ML) models. Compared to absorption-derived parameters, commonly used fluorescence indices such as the fluorescence index (FI), humification index (HIX), biological index (BIX), and protein-like component improved the prediction accuracy. However, the prediction accuracy was decreased when all optical indices were considered for computation due to increased noise and uncertainty in the models. With the exception of chemical oxygen demand (COD), this study successfully replaced biochemical oxygen demand (BOD), dissolved organic carbon (DOC), and nutrients with selected fluorescence indices, demonstrating relatively analogous performance in either training or testing data, with consistent and good coefficient of determination (R2) values of approximately 0.85 and 0.74, respectively. Among all models considered, ensemble learning models consistently outperformed conventional regression models and artificial neural networks (ANNs). However, there was a trade-off between accuracy and computation efficiency among the ensemble learning models (i.e., Stacking and XGBoost) for algal bloom prediction. Our study offers a glimpse of the potential application of spectroscopic measures to improve accuracy and efficiency in algal bloom prediction, but further work should be carried out in other water bodies to further validate our proposed hypothesis.