Second allogeneic hematopoietic stem cell transplantation with reduced-intensity conditioning and donor changes in relapsed hematological malignancies after the first allogeneic transplant / 中华血液学杂志

Objective: The purpose of this study was to assess the safety and efficacy of a second allogeneic hematopoietic stem cell transplantation (allo-HSCT) with reduced-intensity conditioning (RIC) in patients with hematological malignancies who had relapsed after the first allo-HSCT. Methods: Between April 2018 and June 2021, 44 patients with hematological malignancies (B-ALL 23, T-ALL/T-LBL 4, AML15, and MDS 2) were enrolled and retrospectively examined. Unrelated donors (n=12) or haploidentical donors (n=32) were used. Donors were replaced in all patients for the second allo-HSCT. Hematological and immunological germline predisposition genes and hematopoietic and immune function tests were used to select the best-related donor. Total body irradiation (TBI) /fludarabine (FLU) -based (n=38), busulfan (BU) /FLU-based (n=4), total marrow irradiation (TMI) /FLU-based (n=1), and BU/cladribine-based (n=1) were the RIC regimens used. For graft versus host disease (GVHD) prevention, cyclosporine, mycophenolate mofetil, short-term methotrexate, and ATG were used. Eighteen (40.9%) of 44 patients with gene variations for which targeted medications are available underwent post-transplant maintenance therapy. Results: The median age was 25 years old (range: 7-55). The median interval between the first and second HSCT was 19.5 months (range: 6-77). Before the second allo-HSCT, 33 (75%) of the patients were in complete remission (CR), whereas 11 (25%) were not. All patients had long-term engraftment. The grade Ⅱ-Ⅳ GVHD and severe acute GVHD rates were 20.5% and 9.1%, respectively. Chronic GVHD was found in 20.5% of limited patterns and 22.7% of severe patterns. CMV and EBV reactivation rates were 29.5% and 6.8%, respectively. Hemorrhage cystitis occurred in 15.9% of cases, grade Ⅰ or Ⅱ. The 1-yr disease-free survival (DFS), overall survival (OS), and cumulative recurrence incidence (RI) rates of all patients were 72.5% (95% CI, 54.5%-84.3%), 80.6% (95% CI, 63.4%-90.3%), and 25.1% (95% CI, 13.7%-43.2%), respectively, with a median follow-up of 14 (2-39) months. There were eight deaths (seven relapses and one infection). The rate of non-relapse mortality (NRM) was only 2.3%. The CR patients' 1-yr RI rate was significantly lower than the NR patients (16.8% vs 48.1%, P=0.026). The DFS rate in CR patients was greater than in NR patients, although there was no statistical difference (79.9% vs 51.9%, P=0.072). Univariate analysis revealed that CR before the second allo-HSCT was an important prognostic factor. Conclusion: With our RIC regimens, donor change, and post-transplant maintenance therapy, the second allo-HSCT in relapsed hematological malignancies after the first allo-HSCT is a safe and effective treatment with high OS and DFS and low NRM and relapse rate. The most important factor influencing the prognosis of the second allo-HSCT is the patient's illness condition before the transplant.

GaitForeMer: Self-Supervised Pre-Training of Transformers via Human Motion Forecasting for Few-Shot Gait Impairment Severity Estimation.

Parkinson's disease (PD) is a neurological disorder that has a variety of observable motor-related symptoms such as slow movement, tremor, muscular rigidity, and impaired posture. PD is typically diagnosed by evaluating the severity of motor impairments according to scoring systems such as the Movement Disorder Society Unified Parkinson's Disease Rating Scale (MDS-UPDRS). Automated severity prediction using video recordings of individuals provides a promising route for non-intrusive monitoring of motor impairments. However, the limited size of PD gait data hinders model ability and clinical potential. Because of this clinical data scarcity and inspired by the recent advances in self-supervised large-scale language models like GPT-3, we use human motion forecasting as an effective self-supervised pre-training task for the estimation of motor impairment severity. We introduce GaitForeMer, Gait Forecasting and impairment estimation transforMer, which is first pre-trained on public datasets to forecast gait movements and then applied to clinical data to predict MDS-UPDRS gait impairment severity. Our method outperforms previous approaches that rely solely on clinical data by a large margin, achieving an F1 score of 0.76, precision of 0.79, and recall of 0.75. Using GaitForeMer, we show how public human movement data repositories can assist clinical use cases through learning universal motion representations. The code is available at https://github.com/markendo/GaitForeMer.

Socially situated artificial intelligence enables learning from human interaction.

Regardless of how much data artificial intelligence agents have available, agents will inevitably encounter previously unseen situations in real-world deployments. Reacting to novel situations by acquiring new information from other people-socially situated learning-is a core faculty of human development. Unfortunately, socially situated learning remains an open challenge for artificial intelligence agents because they must learn how to interact with people to seek out the information that they lack. In this article, we formalize the task of socially situated artificial intelligence-agents that seek out new information through social interactions with people-as a reinforcement learning problem where the agent learns to identify meaningful and informative questions via rewards observed through social interaction. We manifest our framework as an interactive agent that learns how to ask natural language questions about photos as it broadens its visual intelligence on a large photo-sharing social network. Unlike active-learning methods, which implicitly assume that humans are oracles willing to answer any question, our agent adapts its behavior based on observed norms of which questions people are or are not interested to answer. Through an 8-mo deployment where our agent interacted with 236,000 social media users, our agent improved its performance at recognizing new visual information by 112%. A controlled field experiment confirmed that our agent outperformed an active-learning baseline by 25.6%. This work advances opportunities for continuously improving artificial intelligence (AI) agents that better respect norms in open social environments.

Rethinking Architecture Design for Tackling Data Heterogeneity in Federated Learning.

Federated learning is an emerging research paradigm enabling collaborative training of machine learning models among different organizations while keeping data private at each institution. Despite recent progress, there remain fundamental challenges such as the lack of convergence and the potential for catastrophic forgetting across real-world heterogeneous devices. In this paper, we demonstrate that self-attention-based architectures (e.g., Transformers) are more robust to distribution shifts and hence improve federated learning over heterogeneous data. Concretely, we conduct the first rigorous empirical investigation of different neural architectures across a range of federated algorithms, real-world benchmarks, and heterogeneous data splits. Our experiments show that simply replacing convolutional networks with Transformers can greatly reduce catastrophic forgetting of previous devices, accelerate convergence, and reach a better global model, especially when dealing with heterogeneous data. We release our code and pretrained models to encourage future exploration in robust architectures as an alternative to current research efforts on the optimization front.

Metadata Normalization.

Batch Normalization (BN) and its variants have delivered tremendous success in combating the covariate shift induced by the training step of deep learning methods. While these techniques normalize feature distributions by standardizing with batch statistics, they do not correct the influence on features from extraneous variables or multiple distributions. Such extra variables, referred to as metadata here, may create bias or confounding effects (e.g., race when classifying gender from face images). We introduce the Metadata Normalization (MDN) layer, a new batch-level operation which can be used end-to-end within the training framework, to correct the influence of metadata on feature distributions. MDN adopts a regression analysis technique traditionally used for preprocessing to remove (regress out) the metadata effects on model features during training. We utilize a metric based on distance correlation to quantify the distribution bias from the metadata and demonstrate that our method successfully removes metadata effects on four diverse settings: one synthetic, one 2D image, one video, and one 3D medical image dataset.

Embodied intelligence via learning and evolution.

The intertwined processes of learning and evolution in complex environmental niches have resulted in a remarkable diversity of morphological forms. Moreover, many aspects of animal intelligence are deeply embodied in these evolved morphologies. However, the principles governing relations between environmental complexity, evolved morphology, and the learnability of intelligent control, remain elusive, because performing large-scale in silico experiments on evolution and learning is challenging. Here, we introduce Deep Evolutionary Reinforcement Learning (DERL): a computational framework which can evolve diverse agent morphologies to learn challenging locomotion and manipulation tasks in complex environments. Leveraging DERL we demonstrate several relations between environmental complexity, morphological intelligence and the learnability of control. First, environmental complexity fosters the evolution of morphological intelligence as quantified by the ability of a morphology to facilitate the learning of novel tasks. Second, we demonstrate a morphological Baldwin effect i.e., in our simulations evolution rapidly selects morphologies that learn faster, thereby enabling behaviors learned late in the lifetime of early ancestors to be expressed early in the descendants lifetime. Third, we suggest a mechanistic basis for the above relationships through the evolution of morphologies that are more physically stable and energy efficient, and can therefore facilitate learning and control.

Representation Learning with Statistical Independence to Mitigate Bias.

Presence of bias (in datasets or tasks) is inarguably one of the most critical challenges in machine learning applications that has alluded to pivotal debates in recent years. Such challenges range from spurious associations between variables in medical studies to the bias of race in gender or face recognition systems. Controlling for all types of biases in the dataset curation stage is cumbersome and sometimes impossible. The alternative is to use the available data and build models incorporating fair representation learning. In this paper, we propose such a model based on adversarial training with two competing objectives to learn features that have (1) maximum discriminative power with respect to the task and (2) minimal statistical mean dependence with the protected (bias) variable(s). Our approach does so by incorporating a new adversarial loss function that encourages a vanished correlation between the bias and the learned features. We apply our method to synthetic data, medical images (containing task bias), and a dataset for gender classification (containing dataset bias). Our results show that the learned features by our method not only result in superior prediction performance but also are unbiased.

Quantifying Parkinson's disease motor severity under uncertainty using MDS-UPDRS videos.

Parkinson's disease (PD) is a brain disorder that primarily affects motor function, leading to slow movement, tremor, and stiffness, as well as postural instability and difficulty with walking/balance. The severity of PD motor impairments is clinically assessed by part III of the Movement Disorder Society Unified Parkinson's Disease Rating Scale (MDS-UPDRS), a universally-accepted rating scale. However, experts often disagree on the exact scoring of individuals. In the presence of label noise, training a machine learning model using only scores from a single rater may introduce bias, while training models with multiple noisy ratings is a challenging task due to the inter-rater variabilities. In this paper, we introduce an ordinal focal neural network to estimate the MDS-UPDRS scores from input videos, to leverage the ordinal nature of MDS-UPDRS scores and combat class imbalance. To handle multiple noisy labels per exam, the training of the network is regularized via rater confusion estimation (RCE), which encodes the rating habits and skills of raters via a confusion matrix. We apply our pipeline to estimate MDS-UPDRS test scores from their video recordings including gait (with multiple Raters, R=3) and finger tapping scores (single rater). On a sizable clinical dataset for the gait test (N=55), we obtained a classification accuracy of 72% with majority vote as ground-truth, and an accuracy of ∼84% of our model predicting at least one of the raters' scores. Our work demonstrates how computer-assisted technologies can be used to track patients and their motor impairments, even when there is uncertainty in the clinical ratings. The latest version of the code will be available at https://github.com/mlu355/PD-Motor-Severity-Estimation.

Ethical issues in using ambient intelligence in health-care settings.

Ambient intelligence is increasingly finding applications in health-care settings, such as helping to ensure clinician and patient safety by monitoring staff compliance with clinical best practices or relieving staff of burdensome documentation tasks. Ambient intelligence involves using contactless sensors and contact-based wearable devices embedded in health-care settings to collect data (eg, imaging data of physical spaces, audio data, or body temperature), coupled with machine learning algorithms to efficiently and effectively interpret these data. Despite the promise of ambient intelligence to improve quality of care, the continuous collection of large amounts of sensor data in health-care settings presents ethical challenges, particularly in terms of privacy, data management, bias and fairness, and informed consent. Navigating these ethical issues is crucial not only for the success of individual uses, but for acceptance of the field as a whole.

Mechanism of Suoquanwan on Enuresis Based on Network Pharmacology and Experimental Verification / 中国实验方剂学杂志

Objective:To explore the targets and relevant signaling pathways of Suoquanwan in the treatment of enuresis using network pharmacology，and animal expriments are applied to further define its mechanism of action. Method:Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform（TCMSP） database was used to screen out active chemical components of Suoquanwan，varieties of systematic biological databases were integrated to construct the "active component-disease-target" network relationship，and the common protein protein interaction network（PPI） network genes were functionally enriched. Quantitative real time polymerase chain reaction（Real-time PCR） and Western blot were used to verify the effect of Suoquanwan on AVPR2 and DRD2 gene. Result:A total of 32 active ingredients were screened from Suoquanwan. These active ingredients were interacted with 131 potential targets relating to Enuresis，which contained 14 core target genes，namely arginine vasopressin receptor 2 （AVPR2）， neurotrophic receptor tyrosine kinase 1（NTRK1）， dopamine receptor D2（DRD2）， opioid receptor mu 1（OPRM1）， 5-hydroxytryptamine receptor 1A（HTR1A）， 5-hydroxytryptamine receptor 1B（HTR1B），solute carrier family 6 member 4（SLC6A4），Adrenoceptor Alpha 2A（ADRA2A）， prostaglandin-endoperoxide synthase 2（PTGS2）， cholinergic receptor muscarinic 2（CHRM2）， solute carrier family 6 member 3 （SLC6A3）， 5-hydroxytryptamine receptor 6（HTR6）， solute carrier family 6 member 2（SLC6A2）， cytochrome P450 family 2 subfamily C member 19（CYP2C19）. Gene enrichments mainly involved to G protein-coupled receptor signaling pathway，regulation of trans-synaptic signaling，regulation of neurotransmitter transport and neuroactive ligand-receptor interaction. Real-time PCR and Western blot results showed that Suoquanwan could enhance the expression of AVPR2 in rat kidney，and weaken the expression of DRD2 in rat adrenal. Conclusion:The main chemical constituents in Suoquanwan may alleviate enuresis by regulating AVPR2 and DRD2 and then participating in the G protein-coupled receptor signaling pathway，regulation of trans-synaptic signaling，regulation of neurotransmitter transport and other biological processes.

The mechanism of toxicity enhancement with the combination of Flos Genkwa and Radix et Rhizoma Glycyrrhizae based on the solubilization of glycyrrhizic acid / 药学学报

In order to study the contraindications of the compatibility of Flos Genkwa-Radix et Rhizoma Glycyrrhizae, in this study, the solubilizing and poisoning essence were explored. In this experiment, chromatographic assay, field emission scanning electron microscopy, MTT cytotoxicity evaluation, and other methods were used to study the main chemical components, morphology and toxicity of the ethyl acetate part of Flos Genkwa and its co-decoction with glycyrrhizic acid, in order to clarify Flos Genkwa-Radix et Rhizoma Glycyrrhizae incompatibility provides a new idea for the research on incompatibility of Flos Genkwa-Radix et Rhizoma Glycyrrhizae. The results showed that after co-decoction of the ethyl acetate part of Flos Genkwa with glycyrrhizic acid, high performance liquid chromatography (HPLC) detected the dissolution of the toxic component yuanhuacine of 54.8%, while yuanhuacine chromatographic peak was not detected in the Flos Genkwa ethyl acetate part of the single decoction. The increase of co-decoction dissolution rate was observed by scanning electron microscopy, and it was found that glycyrrhizic acid uniformly dispersed the fat-soluble components of Flos Genkwa into nano-scale particles, which improved the solubility and stability in the solution. Furthermore, the results of cytotoxicity evaluation showed that the survival rate of cells decreased after co-decoction, 4',6-diamidino-2-phenylindole (DAPI) staining also gave the same results. In summary, the co-decoction of the ethyl acetate part of Flos Genkwa with glycyrrhizic acid promotes the dissolution of the toxic component yuanhuacine, and makes the part form uniformly distributed nanoparticles, which is conducive to the absorption of the ingredient and increases the toxicity.

Vision-based Estimation of MDS-UPDRS Gait Scores for Assessing Parkinson's Disease Motor Severity.

Parkinson's disease (PD) is a progressive neurological disorder primarily affecting motor function resulting in tremor at rest, rigidity, bradykinesia, and postural instability. The physical severity of PD impairments can be quantified through the Movement Disorder Society Unified Parkinson's Disease Rating Scale (MDS-UPDRS), a widely used clinical rating scale. Accurate and quantitative assessment of disease progression is critical to developing a treatment that slows or stops further advancement of the disease. Prior work has mainly focused on dopamine transport neuroimaging for diagnosis or costly and intrusive wearables evaluating motor impairments. For the first time, we propose a computer vision-based model that observes non-intrusive video recordings of individuals, extracts their 3D body skeletons, tracks them through time, and classifies the movements according to the MDS-UPDRS gait scores. Experimental results show that our proposed method performs significantly better than chance and competing methods with an F 1-score of 0.83 and a balanced accuracy of 81%. This is the first benchmark for classifying PD patients based on MDS-UPDRS gait severity and could be an objective biomarker for disease severity. Our work demonstrates how computer-assisted technologies can be used to non-intrusively monitor patients and their motor impairments. The code is available at https://github.com/mlu355/PD-Motor-Severity-Estimation.

Illuminating the dark spaces of healthcare with ambient intelligence.

Advances in machine learning and contactless sensors have given rise to ambient intelligence-physical spaces that are sensitive and responsive to the presence of humans. Here we review how this technology could improve our understanding of the metaphorically dark, unobserved spaces of healthcare. In hospital spaces, early applications could soon enable more efficient clinical workflows and improved patient safety in intensive care units and operating rooms. In daily living spaces, ambient intelligence could prolong the independence of older individuals and improve the management of individuals with a chronic disease by understanding everyday behaviour. Similar to other technologies, transformation into clinical applications at scale must overcome challenges such as rigorous clinical validation, appropriate data privacy and model transparency. Thoughtful use of this technology would enable us to understand the complex interplay between the physical environment and health-critical human behaviours.

Automatic detection of hand hygiene using computer vision technology.

OBJECTIVE: Hand hygiene is essential for preventing hospital-acquired infections but is difficult to accurately track. The gold-standard (human auditors) is insufficient for assessing true overall compliance. Computer vision technology has the ability to perform more accurate appraisals. Our primary objective was to evaluate if a computer vision algorithm could accurately observe hand hygiene dispenser use in images captured by depth sensors. MATERIALS AND METHODS: Sixteen depth sensors were installed on one hospital unit. Images were collected continuously from March to August 2017. Utilizing a convolutional neural network, a machine learning algorithm was trained to detect hand hygiene dispenser use in the images. The algorithm's accuracy was then compared with simultaneous in-person observations of hand hygiene dispenser usage. Concordance rate between human observation and algorithm's assessment was calculated. Ground truth was established by blinded annotation of the entire image set. Sensitivity and specificity were calculated for both human and machine-level observation. RESULTS: A concordance rate of 96.8% was observed between human and algorithm (kappa = 0.85). Concordance among the 3 independent auditors to establish ground truth was 95.4% (Fleiss's kappa = 0.87). Sensitivity and specificity of the machine learning algorithm were 92.1% and 98.3%, respectively. Human observations showed sensitivity and specificity of 85.2% and 99.4%, respectively. CONCLUSIONS: A computer vision algorithm was equivalent to human observation in detecting hand hygiene dispenser use. Computer vision monitoring has the potential to provide a more complete appraisal of hand hygiene activity in hospitals than the current gold-standard given its ability for continuous coverage of a unit in space and time.

Online Developmental Science to Foster Innovation, Access, and Impact.

We propose that developmental cognitive science should invest in an online CRADLE, a Collaboration for Reproducible and Distributed Large-Scale Experiments that crowdsources data from families participating on the internet. Here, we discuss how the field can work together to further expand and unify current prototypes for the benefit of researchers, science, and society.

Assessing the accuracy of automatic speech recognition for psychotherapy.

Accurate transcription of audio recordings in psychotherapy would improve therapy effectiveness, clinician training, and safety monitoring. Although automatic speech recognition software is commercially available, its accuracy in mental health settings has not been well described. It is unclear which metrics and thresholds are appropriate for different clinical use cases, which may range from population descriptions to individual safety monitoring. Here we show that automatic speech recognition is feasible in psychotherapy, but further improvements in accuracy are needed before widespread use. Our HIPAA-compliant automatic speech recognition system demonstrated a transcription word error rate of 25%. For depression-related utterances, sensitivity was 80% and positive predictive value was 83%. For clinician-identified harm-related sentences, the word error rate was 34%. These results suggest that automatic speech recognition may support understanding of language patterns and subgroup variation in existing treatments but may not be ready for individual-level safety surveillance.

Active constituents and mechanisms of Respiratory Detox Shot, a traditional Chinese medicine prescription, for COVID-19 control and prevention: Network-molecular docking-LC-MS analysis / 中西医结合学报

OBJECTIVE@#Lung-toxin Dispelling Formula No. 1, referred to as Respiratory Detox Shot (RDS), was developed based on a classical prescription of traditional Chinese medicine (TCM) and the theoretical understanding of herbal properties within TCM. Therapeutic benefits of using RDS for both disease control and prevention, in the effort to contain the coronavirus disease 2019 (COVID-19), have been shown. However, the biochemically active constituents of RDS and their mechanisms of action are still unclear. The goal of the present study is to clarify the material foundation and action mechanism of RDS.@*METHODS@#To conduct an analysis of RDS, an integrative analytical platform was constructed, including target prediction, protein-protein interaction (PPI) network, and cluster analysis; further, the hub genes involved in the disease-related pathways were identified, and the their corresponding compounds were used for in vitro validation of molecular docking predictions. The presence of these validated compounds was also measured in samples of the RDS formula to quantify the abundance of the biochemically active constituents. In our network pharmacological study, a total of 26 bioinformatic programs and databases were used, and six networks, covering the entire Zang-fu viscera, were constructed to comprehensively analyze the intricate connections among the compounds-targets-disease pathways-meridians of RDS.@*RESULTS@#For all 1071 known chemical constituents of the nine ingredients in RDS, identified from established TCM databases, 157 passed drug-likeness screening and led to 339 predicted targets in the constituent-target network. Forty-two hub genes with core regulatory effects were extracted from the PPI network, and 134 compounds and 29 crucial disease pathways were implicated in the target-constituent-disease network. Twelve disease pathways attributed to the Lung-Large Intestine meridians, with six and five attributed to the Kidney-Urinary Bladder and Stomach-Spleen meridians, respectively. One-hundred and eighteen candidate constituents showed a high binding affinity with SARS-coronavirus-2 3-chymotrypsin-like protease (3CL), as indicated by molecular docking using computational pattern recognition. The in vitro activity of 22 chemical constituents of RDS was validated using the 3CL inhibition assay. Finally, using liquid chromatography mass spectrometry in data-independent analysis mode, the presence of seven out of these 22 constituents was confirmed and validated in an aqueous decoction of RDS, using reference standards in both non-targeted and targeted approaches.@*CONCLUSION@#RDS acts primarily in the Lung-Large Intestine, Kidney-Urinary Bladder and Stomach-Spleen meridians, with other Zang-fu viscera strategically covered by all nine ingredients. In the context of TCM meridian theory, the multiple components and targets of RDS contribute to RDS's dual effects of health-strengthening and pathogen-eliminating. This results in general therapeutic effects for early COVID-19 control and prevention.

A computer vision system for deep learning-based detection of patient mobilization activities in the ICU.

Early and frequent patient mobilization substantially mitigates risk for post-intensive care syndrome and long-term functional impairment. We developed and tested computer vision algorithms to detect patient mobilization activities occurring in an adult ICU. Mobility activities were defined as moving the patient into and out of bed, and moving the patient into and out of a chair. A data set of privacy-safe-depth-video images was collected in the Intermountain LDS Hospital ICU, comprising 563 instances of mobility activities and 98,801 total frames of video data from seven wall-mounted depth sensors. In all, 67% of the mobility activity instances were used to train algorithms to detect mobility activity occurrence and duration, and the number of healthcare personnel involved in each activity. The remaining 33% of the mobility instances were used for algorithm evaluation. The algorithm for detecting mobility activities attained a mean specificity of 89.2% and sensitivity of 87.2% over the four activities; the algorithm for quantifying the number of personnel involved attained a mean accuracy of 68.8%.

Scene Graph Prediction with Limited Labels.

Visual knowledge bases such as Visual Genome power numerous applications in computer vision, including visual question answering and captioning, but suffer from sparse, incomplete relationships. All scene graph models to date are limited to training on a small set of visual relationships that have thousands of training labels each. Hiring human annotators is expensive, and using textual knowledge base completion methods are incompatible with visual data. In this paper, we introduce a semi-supervised method that assigns probabilistic relationship labels to a large number of unlabeled images using few' labeled examples. We analyze visual relationships to suggest two types of image-agnostic features that are used to generate noisy heuristics, whose outputs are aggregated using a factor graph-based generative model. With as few as 10 labeled examples per relationship, the generative model creates enough training data to train any existing state-of-the-art scene graph model. We demonstrate that our method outperforms all baseline approaches on scene graph prediction by 5.16 recall@ 100 for PREDCLS. In our limited label setting, we define a complexity metric for relationships that serves as an indicator (R2 = 0.778) for conditions under which our method succeeds over transfer learning, the de-facto approach for training with limited labels.