Ensemble learning for predicting ex vivo human placental barrier permeability.

BACKGROUND: The placental barrier protects the fetus from exposure to some toxicants and is vital for drug development and risk assessment of environmental chemicals. However, in vivo experiments for assessing the placental barrier permeability of chemicals is not ethically acceptable. Although ex vivo placental perfusion methods provide good alternatives for the assessment of placental barrier permeability, the application to a large number of test chemicals could be time- and resource-consuming. Computational prediction models for ex vivo placental barrier permeability are therefore desirable. METHODS: A total of 87 chemicals and corresponding 1444 physicochemical properties were divided into training and test datasets. Three types of algorithms including linear regression, random forest, and ensemble models were applied to develop prediction models for ex vivo placental barrier permeability. RESULTS: Among the tested models, the ensemble model integrating the previous two methods performed best for predicting ex vivo human placental barrier permeability with correlation coefficients of 0.887 and 0.825 when considering the applicability domain. An additional test on seven newly curated chemicals from the literature showed a good correlation coefficient of 0.879 which was further improved to 0.921 by considering the variation of experiments. CONCLUSION: In this study, the first valid predicting model for ex vivo human placental barrier permeability was developed following the OECD guideline. The model is expected to be useful for assessing the human placental barrier permeability and can be integrated with developmental toxicity prediction models for investigating the toxic effects of chemicals on the fetus.

Current trends in read-across applications for chemical risk assessments and chemical registrations in the Republic of Korea.

Read-across, an alternative approach for hazard assessment, has been widely adopted when in vivo data are unavailable for chemicals of interest. Read-across is enabled via in silico tools such as quantitative structure activity relationship (QSAR) modeling. In this study, the current status of structure activity relationship (SAR)-based read-across applications in the Republic of Korea (ROK) was examined considering both chemical risk assessments and chemical registrations from different sectors, including regulatory agencies, industry, and academia. From the regulatory perspective, the Ministry of Environment (MOE) established the Act on Registration and Evaluation of Chemicals (AREC) in 2019 to enable registrants to submit alternative data such as information from read-across instead of in vivo data to support hazard assessment and determine chemical-specific risks. Further, the Ministry of Food and Drug Safety (MFDS) began to consider read-across approaches for establishing acceptable intake (AI) limits of impurities occurring during pharmaceutical manufacturing processes under the ICH M7 guideline. Although read-across has its advantages, this approach also has limitations including (1) lack of standardized criteria for regulatory acceptance, (2) inconsistencies in the robustness of scientific evidence, and (3) deficiencies in the objective reliability of read-across data. The application and acceptance rate of read-across may vary among regulatory agencies. Therefore, sufficient data need to be prepared to verify the hypothesis that structural similarities might lead to similarities in properties of substances (between source and target chemicals) prior to adopting a read-across approach. In some cases, additional tests may be required during the registration process to clarify long-term effects on human health or the environment for certain substances that are data deficient. To improve the quality of read-across data for regulatory acceptance, cooperative efforts from regulatory agencies, academia, and industry are needed to minimize limitations of read-across applications.

Prediction of Synergistic Toxicity of Binary Mixtures to Vibrio fischeri Based on Biomolecular Interaction Networks.

The paradigm of chemical safety assessment is shifting from 'chemical management focusing on single chemicals' to 'product management extending to mixtures and articles'. However, because of the enormous combinatorial complexity, testing the toxicity of all conceivable mixture products is currently not feasible. There exist only few models that allow predicting the synergistic toxicity potentially caused by toxicological interactions among components. In this study, we present a novel approach to qualitatively predict the synergistic toxicity of binary mixtures to Vibrio fischeri. On the basis of information derived from protein-chemical and protein-protein interaction networks, we trained machine learning models for classifying chemical mixtures to have synergistic or nonsynergistic toxicity with accuracies and an area under the receiver operating characteristic (ROC) curve (AUC) up to 0.73. The numbers of shared targets and their neighborhood were found to be the most important features for classifying chemicals into synergistic and nonsynergistic groups.

Quasi-SMILES-Based Nano-Quantitative Structure-Activity Relationship Model to Predict the Cytotoxicity of Multiwalled Carbon Nanotubes to Human Lung Cells.

Quantitative structure-activity relationship (QSAR) models for nanomaterials (nano-QSAR) were developed to predict the cytotoxicity of 20 different types of multiwalled carbon nanotubes (MWCNTs) to human lung cells by using quasi-SMILES. The optimal descriptors, recorded as quasi-SMILES, were encoded to represent the physicochemical properties and experimental conditions for the MWCNTs from 276 data records collected from previously published studies. The quasi-SMILES used to build the optimal descriptors were (i) diameter, (ii) length, (iii) surface area, (iv) in vitro toxicity assay, (v) cell line, (vi) exposure time, and (vii) dose. The model calculations were performed by using the Monte Carlo method and computed with CORAL software ( www.insilico.eu/coral ). The quasi-SMILES-based nano-QSAR model provided satisfactory statistical results ( R2 for internal validation data sets: 0.60-0.80; R2pred for external validation data sets: 0.81-0.88). The model showed potential for use in the estimation of human lung cell viability after exposure to MWCNTs with the following properties: diameter, 12-74 nm; length, 0.19-20.25 µm; surface area, 11.3-380.0 m2/g; and dose, 0-200 ppm.

Flexible multimaterial fibers in modern biomedical applications.

Biomedical devices are indispensable in modern healthcare, significantly enhancing patients' quality of life. Recently, there has been a drastic increase in innovations for the fabrication of biomedical devices. Amongst these fabrication methods, the thermal drawing process has emerged as a versatile and scalable process for the development of advanced biomedical devices. By thermally drawing a macroscopic preform, which is meticulously designed and integrated with functional materials, hundreds of meters of multifunctional fibers are produced. These scalable flexible multifunctional fibers are embedded with functionalities such as electrochemical sensing, drug delivery, light delivery, temperature sensing, chemical sensing, pressure sensing, etc. In this review, we summarize the fabrication method of thermally drawn multifunctional fibers and highlight recent developments in thermally drawn fibers for modern biomedical application, including neural interfacing, chemical sensing, tissue engineering, cancer treatment, soft robotics and smart wearables. Finally, we discuss the existing challenges and future directions of this rapidly growing field.

T-DOpE probes reveal sensitivity of hippocampal oscillations to cannabinoids in behaving mice.

Understanding the neural basis of behavior requires monitoring and manipulating combinations of physiological elements and their interactions in behaving animals. We developed a thermal tapering process enabling fabrication of low-cost, flexible probes combining ultrafine features: dense electrodes, optical waveguides, and microfluidic channels. Furthermore, we developed a semi-automated backend connection allowing scalable assembly. We demonstrate T-DOpE (Tapered Drug delivery, Optical stimulation, and Electrophysiology) probes achieve in single neuron-scale devices (1) high-fidelity electrophysiological recording (2) focal drug delivery and (3) optical stimulation. The device tip can be miniaturized (as small as 50 µm) to minimize tissue damage while the ~20 times larger backend allows for industrial-scale connectorization. T-DOpE probes implanted in mouse hippocampus revealed canonical neuronal activity at the level of local field potentials (LFP) and neural spiking. Taking advantage of the triple-functionality of these probes, we monitored LFP while manipulating cannabinoid receptors (CB1R; microfluidic agonist delivery) and CA1 neuronal activity (optogenetics). Focal infusion of CB1R agonist downregulated theta and sharp wave-ripple oscillations (SPW-Rs). Furthermore, we found that CB1R activation reduces sharp wave-ripples by impairing the innate SPW-R-generating ability of the CA1 circuit.

Granular retrosplenial cortex layer 2/3 generates high-frequency oscillations dynamically coupled with hippocampal rhythms across brain states.

The granular retrosplenial cortex (gRSC) exhibits high-frequency oscillations (HFOs; ∼150 Hz), which can be driven by a hippocampus-subiculum pathway. How the cellular-synaptic and laminar organization of gRSC facilitates HFOs is unknown. Here, we probe gRSC HFO generation and coupling with hippocampal rhythms using focal optogenetics and silicon-probe recordings in behaving mice. ChR2-mediated excitation of CaMKII-expressing cells in L2/3 or L5 induces HFOs, but spontaneous HFOs are found only in L2/3, where HFO power is highest. HFOs couple to CA1 sharp wave-ripples (SPW-Rs) during rest and the descending phase of theta. gRSC HFO current sources and sinks are the same for events during both SPW-Rs and theta oscillations. Independent component analysis shows that high gamma (50-100 Hz) in CA1 stratum lacunosum moleculare is comodulated with HFO power. HFOs may thus facilitate interregional communication of a multisynaptic loop between the gRSC, hippocampus, and medial entorhinal cortex during distinct brain and behavioral states.

Multifunctional Tetrode-like Drug delivery, Optical stimulation, and Electrophysiology (Tetro-DOpE) probes.

Having reliable tools for recording and manipulating circuit activity are essential to understand the complex patterns of neural dynamics that underlie brain function. We present Tetro-DOpE (Tetrode-like Drug delivery, Optical stimulation, and Electrophysiology) probes that can simultaneously record and manipulate neural activity in behaving rodents. We fabricated thin multifunctional fibers (<50 µm) using the scalable convergence thermal drawing process. Then, the thin fibers are bundled, similar to tetrode fabrication, to produce Tetro-DOpE probes. We demonstrated the multifunctionality (i.e., electrophysiology, optical stimulation, and drug delivery) of our probe in head-fixed behaving mice. Furthermore, we assembled a six-shank probe mounted on a microdrive which enabled stable recordings of over months when chronically implanted in freely behaving mice. These in vivo experiments demonstrate the potential of customizable, low cost, and accessible multifunctional Tetro-DOpE probes for investigation of neural circuitry in behaving animals.

A case study and a computational simulation of the European Union draft technical guidance documents for chemical safety assessment of mixtures: limitations and a tentative alternative.

Two different methods, Key Critical Component (KCC) and Composite Reciprocal (CR), are used to derive Predicted No Effect Concentration (PNECs) and Derived No Effect Level (DNELs) that are employed in mixture risk assessment. A case study was conducted on coating products and a computational simulation of the KCC and CR methods. The KCC method considers only one key critical component as the whole mixture of equal danger, whereas the CR method indirectly considers the contribution of each component to mixture toxicity. Computational simulation is the first approach that adopts a theoretical perspective in discussing the causes of different outcomes between models and how the two methods should be applied for regulatory purposes in terms of concept, implementation, and performance. The deviation between the two models was highly dependent on the number of mixture components having similarly weighted DNELs (or PNECs). Therefore, if risk assessment using the KCC or CR method needs to be achieved, then the number of mixture components with similarly weighted DNELs (or PNECs) should be checked first. In this study, as a tentative alternative to the existing KCC and CR methods, a tiered approach combining Enhanced KCC (e-KCC) and CR methods was proposed and discussed.

Artificial Action Potential and Ionic Power Device Inspired by Ion Channels and Excitable Cell.

In vivo, the membrane potential of the excitable cell working by ion gradients plays a significant role in bioelectricity generation and nervous system operation. Conventional bioinspired power systems generally have adopted ion gradients, but overlook the functions of ion channels and Donnan effect to generate efficient ion flow in the cell. Here, cell-inspired ionic power device implementing the Donnan effect using multi-ions and monovalent ion exchange membranes as artificial ion channels is realized. Different ion-rich electrolytes on either side of the selective membrane generate the ion gradient potentials with high ionic currents and reduce the osmotic imbalance of the membrane. Based on this device, the artificial neuronal signaling is presented by the mechanical switching system of the ion selectivity like mechanosensitive ion channels in a sensory neuron. Compared with reverse electrodialysis, which requires a low concentration, a high-power device with ten times the current and 8.5 times the power density is fabricated. This device activates grown muscle cells by increasing power through serial connection like an electric eel, and shows the possibility of an ion-based artificial nervous system.

Tapered Drug delivery, Optical stimulation, and Electrophysiology (T-DOpE) probes reveal the importance of cannabinoid signaling in hippocampal CA1 oscillations in behaving mice.

Understanding the neural basis of behavior requires monitoring and manipulating combinations of physiological elements and their interactions in behaving animals. Here we developed a thermal tapering process (TTP) which enables the fabrication of novel, low-cost, flexible probes that combine ultrafine features of dense electrodes, optical waveguides, and microfluidic channels. Furthermore, we developed a semi-automated backend connection allowing scalable assembly of the probes. We demonstrate that our T-DOpE ( T apered D rug delivery, Op tical stimulation, and E lectrophysiology) probe achieves in a single neuron-scale device (1) high-fidelity electrophysiological recording (2) focal drug delivery and (3) optical stimulation. With a tapered geometry, the device tip can be minimized (as small as 50 µm) to ensure minimal tissue damage while the backend is ~20 times larger allowing for direct integration with industrial-scale connectorization. Acute and chronic implantation of the probes in mouse hippocampus CA1 revealed canonical neuronal activity at the level of local field potentials and spiking. Taking advantage of the triple-functionality of the T-DOpE probe, we monitored local field potentials with simultaneous manipulation of endogenous type 1 cannabinoid receptors (CB1R; via microfluidic agonist delivery) and CA1 pyramidal cell membrane potential (optogenetic activation). Electro-pharmacological experiments revealed that focal infusion of CB1R agonist CP-55,940 in dorsal CA1 downregulated theta and sharp wave-ripple oscillations. Furthermore, using the full electro-pharmacological-optical feature set of the T-DOpE probe we found that CB1R activation reduces sharp wave-ripples (SPW-Rs) by impairing the innate SPW-R-generating ability of the CA1 circuit.

Multifunctional Fiber-Based Optoacoustic Emitter as a Bidirectional Brain Interface.

A bidirectional brain interface with both "write" and "read" functions can be an important tool for fundamental studies and potential clinical treatments for neurological diseases. Herein, a miniaturized multifunctional fiber-based optoacoustic emitter (mFOE) is reported thatintegrates simultaneous optoacoustic stimulation for "write" and electrophysiology recording of neural circuits for "read". Because of the intrinsic ability of neurons to respond to acoustic wave, there is no requirement of the viral transfection. The orthogonality between optoacoustic waves and electrical field provides a solution to avoid the interference between electrical stimulation and recording. The stimulation function of the mFOE is first validated in cultured ratcortical neurons using calcium imaging. In vivo application of mFOE for successful simultaneous optoacoustic stimulation and electrical recording of brain activities is confirmed in mouse hippocampus in both acute and chronical applications up to 1 month. Minor brain tissue damage is confirmed after these applications. The capability of simultaneous neural stimulation and recording enabled by mFOE opens up new possibilities for the investigation of neural circuits and brings new insights into the study of ultrasound neurostimulation.

Multifunctional fiber-based optoacoustic emitter for non-genetic bidirectional neural communication.

A bidirectional brain interface with both "write" and "read" functions can be an important tool for fundamental studies and potential clinical treatments for neurological diseases. Here we report a miniaturized multifunctional fiber based optoacoustic emitter (mFOE) that first integrates simultaneous non-genetic optoacoustic stimulation for "write" and electrophysiology recording of neural circuits for "read". The non-genetic feature addresses the challenges of the viral transfection required by optogenetics in primates and human. The orthogonality between optoacoustic waves and electrical field provides a solution to avoid the interference between electrical stimulation and recording. We first validated the non-genetic stimulation function of the mFOE in rat cultured neurons using calcium imaging. In vivo application of mFOE for successful simultaneous optoacoustic stimulation and electrical recording of brain activities was confirmed in mouse hippocampus in both acute and chronical applications up to 1 month. Minimal brain tissue damage has been confirmed after these applications. The capability of non-genetic neural stimulation and recording enabled by mFOE opens up new possibilities for the investigation of neural circuits and brings new insights into the study of ultrasound neurostimulation.

Granular retrosplenial cortex layer 2/3 generates high-frequency oscillations coupled with hippocampal theta and gamma in online states or sharp-wave ripples in offline states.

Neuronal oscillations support information transfer by temporally aligning the activity of anatomically distributed 'writer' and 'reader' cell assemblies. High-frequency oscillations (HFOs) such as hippocampal CA1 sharp-wave ripples (SWRs; 100-250 Hz) are sufficiently fast to initiate synaptic plasticity between assemblies and are required for memory consolidation. HFOs are observed in parietal and midline cortices including granular retrosplenial cortex (gRSC). In 'offline' brain states (e.g. quiet wakefulness) gRSC HFOs co-occur with CA1 SWRs, while in 'online' states (e.g. ambulation) HFOs persist with the emergence of theta oscillations. The mechanisms of gRSC HFO oscillations, specifically whether the gRSC can intrinsically generate HFOs, and which layers support HFOs across states, remain unclear. We addressed these issues in behaving mice using optogenetic excitation in individual layers of the gRSC and high density silicon-probe recordings across gRSC layers and hippocampus CA1. Optogenetically induced HFOs (iHFOs) could be elicited by depolarizing excitatory neurons with 100 ms half-sine wave pulses in layer 2/3 (L2/3) or layer 5 (L5) though L5 iHFOs were of lower power than in L2/3. Critically, spontaneous HFOs were only observed in L2/3 and never in L5. Intra-laminar monosynaptic connectivity between excitatory and inhibitory neurons was similar across layers, suggesting other factors restrict HFOs to L2/3. To compare HFOs in online versus offline states we analyzed, separately, HFOs that did or did not co-occur with CA1 SWRs. Using current-source density analysis we found uniform synaptic inputs to L2/3 during all gRSC HFOs, suggesting layer-specific inputs may dictate the localization of HFOs to L2/3. HFOs occurring without SWRs were aligned with the descending phase of both gRSC and CA1 theta oscillations and were coherent with CA1 high frequency gamma oscillations (50-80 Hz). These results demonstrate that gRSC can internally generate HFOs without rhythmic inputs and that HFOs occur exclusively in L2/3, coupled to distinct hippocampal oscillations in online versus offline states.

Submillimeter Multifunctional Ferromagnetic Fiber Robots for Navigation, Sensing, and Modulation.

Small-scale robots capable of remote active steering and navigation offer great potential for biomedical applications. However, the current design and manufacturing procedure impede their miniaturization and integration of various diagnostic and therapeutic functionalities. Herein, submillimeter fiber robots that can integrate navigation, sensing, and modulation functions are presented. These fiber robots are fabricated through a scalable thermal drawing process at a speed of 4 meters per minute, which enables the integration of ferromagnetic, electrical, optical, and microfluidic composite with an overall diameter of as small as 250 µm and a length of as long as 150 m. The fiber tip deflection angle can reach up to 54o under a uniform magnetic field of 45 mT. These fiber robots can navigate through complex and constrained environments, such as artificial vessels and brain phantoms. Moreover, Langendorff mouse hearts model, glioblastoma micro platforms, and in vivo mouse models are utilized to demonstrate the capabilities of sensing electrophysiology signals and performing a localized treatment. Additionally, it is demonstrated that the fiber robots can serve as endoscopes with embedded waveguides. These fiber robots provide a versatile platform for targeted multimodal detection and treatment at hard-to-reach locations in a minimally invasive and remotely controllable manner.

Multifunctional ferromagnetic fiber robots for navigation, sensing, and treatment in minimally invasive surgery.

Small-scale robots capable of remote active steering and navigation offer great potential for biomedical applications. However, the current design and manufacturing procedure impede their miniaturization and integration of various diagnostic and therapeutic functionalities. Here, we present a robotic fiber platform for integrating navigation, sensing, and therapeutic functions at a submillimeter scale. These fiber robots consist of ferromagnetic, electrical, optical, and microfluidic components, fabricated with a thermal drawing process. Under magnetic actuation, they can navigate through complex and constrained environments, such as artificial vessels and brain phantoms. Moreover, we utilize Langendorff mouse hearts model, glioblastoma microplatforms, and in vivo mouse models to demonstrate the capabilities of sensing electrophysiology signals and performing localized treatment. Additionally, we demonstrate that the fiber robots can serve as endoscopes with embedded waveguides. These fiber robots provide a versatile platform for targeted multimodal detection and treatment at hard-to-reach locations in a minimally invasive and remotely controllable manner.

MRA Toolbox v. 1.0: a web-based toolbox for predicting mixture toxicity of chemical substances in chemical products.

The chemical risk assessment paradigm is shifting from "substance-based" to "product/mixture-based" and from "animal testing" to "alternative testing" under chemical regulations. Organisms and the environment may be exposed to mixtures rather than a single substance. Conducting toxicity tests for all possible combinations is impractical due to the enormous combinatorial complexity. This study highlights the development and application case studies of Mixture Risk Assessment Toolbox, a novel web-based platform that supports mixture risk assessment through the use of different prediction models and public databases. This integrated framework provides new functional values for assessors to easily screen and compare the toxicity of mixture products using different computational techniques and find strategic solutions to reduce the mixture toxicity in the product development process. The toolbox ( https://www.mratoolbox.org ) includes four additive toxicity models: two conventional (Concentration Addition; and Independent Action) and two advanced (Generalized Concentration Addition; and Quantitative Structure-Activity Relationship-based Two-Stage Prediction) models. We demonstrated the multiple functions of the toolbox using three cases: (i) how it can be used to calculate the mixture toxicity, (ii) those for which safety data sheet (SDS) only indicating representative toxicity values (EC50; and LC50), and (iii) those comprising chemicals with low toxic effects.

Developing random forest based QSAR models for predicting the mixture toxicity of TiO2 based nano-mixtures to Daphnia magna.

During emission, TiO2 nanoparticles (NPs) might meet various chemicals, including metal ions and organic compounds in aquatic environments (e.g., surface water, sediments). At environmentally safe concentrations, combinations of both TiO2 NPs and those chemicals might cause cocktail effects (i.e., mixture toxicity) to aquatic organisms. Previous models such as concentration addition and independent action require dose-response curves of single components in the mixtures to predict the mixture toxicity. Structure-activity relationship (QSAR) models might predict the toxicity of nano-mixtures without dose-response curves of single components in the mixtures. However, current quantitative structure-activity relationship (QSAR) models are mainly focused on predicting cytotoxicity (i.e., cell viability) of heterogeneous metallic TiO2 nanoparticles (NPs) or mixtures of TiO2 NPs and four metal ions (Cu2+, Cd2+, Ni2+, and Zn2+). To minimize the experimental cost of nano-mixture risk assessment, in this study, we developed novel nano-mixture QSAR models to predict i) EC50 of 76 nano-mixtures containing TiO2 NPs and one of eight inorganic/organic compounds (i.e., AgNO3, Cd(NO3)2, Cu(NO3)2, CuSO4, Na2HAsO4, NaAsO2, Benzylparaben and Benzophenone-3), to Daphnia magna(D. magna), and ii) immobilization of D. magna exposed to one of 98 mixtures containing TiO2 NPs and one of eleven inorganic/organic compounds (i.e., AgNO3, Cd(NO3)2, Cu(NO3)2, CuSO4, Na2HAsO4, NaAsO2, Benzylparaben Benzophenone-3, Pirimicarb, Pentabromodiphenyl Ether and Triton X-100). The nano-mixture QSAR models were developed with mixture descriptors (Dmix) combing quantum descriptors of mixture components (e.g., TiO2 NPs and its partners) by using different machine learning techniques (i.e., random forest, neural network, support vector machine, and multiple linear regression). Nano-mixture QSAR models built with the random forest algorithm and proposed mixture descriptors exhibited good performance for predicting logEC50 (Adj.R2test = 0.955 ± 0.003, RMSEtest = 0.016 ± 0.002, and MAEtest = 0.008 ± 0.001) and immobilization (Adj.R2test = 0.888 ± 0.011, RMSEtest = 11.327 ± 0.730, and MAEtest = 5.933 ± 0.442). The models developed in this study were implemented in a user-friendly application for assessing the aquatic toxicity of TiO2 based nano-mixtures.

Laser Machined Fiber-based Microprobe: Application in Microscale Electroporation.

Microscale electroporation devices are mostly restricted to in vitro experiments (i.e., microchannel and microcapillary). Novel fiber-based microprobes can enable in vivo microscale electroporation and arbitrarily select the cell groups of interest to electroporate. We developed a flexible, fiber-based microscale electroporation device through a thermal drawing process and femtosecond laser micromachining techniques. The fiber consists of four copper electrodes (80 µm), one microfluidic channel (30 µm), and has an overall diameter of 400 µm. The dimensions of the exposed electrodes and channel were customizable through a delicate femtosecond laser setup. The feasibility of the fiber probe was validated through numerical simulations and in vitro experiments. Successful reversible and irreversible microscale electroporation was observed in a 3D collagen scaffold (seeded with U251 human glioma cells) using fluorescent staining. The ablation regions were estimated by performing the covariance error ellipse method and compared with the numerical simulations. The computational and experimental results of the working fiber-based microprobe suggest the feasibility of in vivo microscale electroporation in space-sensitive areas, such as the deep brain.

Status Quo in Data Availability and Predictive Models of Nano-Mixture Toxicity.

Co-exposure of nanomaterials and chemicals can cause mixture toxicity effects to living organisms. Predictive models might help to reduce the intensive laboratory experiments required for determining the toxicity of the mixtures. Previously, concentration addition (CA), independent action (IA), and quantitative structure-activity relationship (QSAR)-based models were successfully applied to mixtures of organic chemicals. However, there were few studies concerning predictive models for toxicity of nano-mixtures before June 2020. Previous reviews provided comprehensive knowledge of computational models and mechanisms for chemical mixture toxicity. There is a gap in the reviewing of datasets and predictive models, which might cause obstacles in the toxicity assessment of nano-mixtures by using in silico approach. In this review, we collected 183 studies of nano-mixture toxicity and curated data to investigate the current data and model availability and gap and to derive research challenges to facilitate further experimental studies for data gap filling and the development of predictive models.