Transfer learning is a crucial technique for handling a small amount of data that is potentially related to other abundant data. However, most of the existing methods are focused on classification tasks using images and language datasets. Therefore, in order to expand the transfer learning scheme to regression tasks, we propose a novel transfer technique based on differential geometry, namely the Geometrically Aligned Transfer Encoder (GATE). In this method, we interpret the latent vectors from the model to exist on a Riemannian curved manifold. We find a proper diffeomorphism between pairs of tasks to ensure that every arbitrary point maps to a locally flat coordinate in the overlapping region, allowing the transfer of knowledge from the source to the target data. This also serves as an effective regularizer for the model to behave in extrapolation regions. In this article, we demonstrate that GATE outperforms conventional methods and exhibits stable behavior in both the latent space and extrapolation regions for various molecular graph datasets.

pdf/geometrically-aligned-transfer-encoder-for-inductive-12ic7d85ye.pdf

Geometrically Aligned Transfer Encoder for Inductive Transfer in Regression Tasks

PURPOSE
We investigated both the clinical utilities and the prognostic impacts of the clonotypic peptide mass spectrometry (MS)-EasyM, a blood-based minimal residual disease (MRD) monitoring protocol in multiple myeloma (MM).


PATIENTS AND METHODS
447 sequential serum samples from 56 MM patients were analyzed using EasyM. Patient-specific M-protein peptides were sequenced from diagnostic sample; sequential samples were quantified by EasyM to monitor the M-protein. The performance of EasyM was compared with serum immunofixation electrophoresis (IFE), bone marrow multiparameter flow cytometry (MFC), and next-generation flow cytometry (NGF) detection. The optimal balance of EasyM sensitivity/specificity versus NGF (10-5 sensitivity) was determined and the prognostic impact of MS-MRD status was investigated.


RESULTS
Of the 447 serum samples detected and measured by EasyM, 397, 126, and 92 had time-matching results for comparison with serum IFE, MFC-MRD, and NGF-MRD, respectively. Using a dotp >0.9 as the MS-MRD positive , sensitivity was 99.6% vs. IFE and 100.0% vs. MFC and NGF. Using an MS negative cutoff informed by ROC analysis (<1.86% of that at diagnosis), EasyM sensitivity remained high versus IFE (88.3%), MFC (85.1%), and NGF (93.2%), while specificity increased to 90.4%, 55.8%, and 93.2%, respectively. In the multivariate analysis, older diagnostic age was an independent predictor for PFS (HR: 3.15 [1.26, 7.86]), the best MS-MRD status (MS-MRD negative) was independent predictor for both PFS (HR: 0.25 [0.12, 0.52]) and OS (HR: 0.16 [0.06, 0.40]).


CONCLUSIONS
EasyM is a highly sensitive and minimal invasive method of MRD monitoring in MM, MS-MRD had significant predictive ability for survival outcomes.

Monitoring minimal residual disease in patients with multiple myeloma by targeted tracking serum M-protein using mass spectrometry (EasyM).

<title>Abstract</title> Minimal Residual Disease (MRD) assessment is a known surrogate marker for survival in multiple myeloma (MM). Here, we present a single institution’s experience assessing MRD by NGS of Ig genes and the long-term impact of depth of response as well as clonal diversity on the clinical outcome of a large population of MM patients; 482 MM patients at the University of California, San Francisco (UCSF) diagnosed from 2008 to 2020 were analyzed retrospectively. MRD assessment was performed by NGS. PFS curves were plotted by the Kaplan-Meier method. In the newly diagnosed group, 119 of 304, achieved MRD negativity at the level of 10− 6 at least once. These patients had a prolonged PFS versus patients who were persistently MRD positive at different levels (p &gt; 0.0001). In the relapsed disease group, 64 of 178 achieved MRD negativity at 10− 6 and PFS was prolonged versus patients who remained MRD positive (p = 0.03). Three categories of MRD dynamics were defined by artificial intelligence: (A) patients with ≥ 3 consistently MRD negative samples, (B) patients with continuously declining but detectable clones, (C) patients with either increasing or a stable number of clones. Groups A and B had a more prolonged PFS than group C (p &lt; 10− 7). Patients who were MRD positive and had not yet relapsed had a higher clonal diversity than those patients who were MRD positive and had relapsed. MRD dynamics can accurately predict disease evolution and drive clinical decision-making. Clonal Diversity could complement MRD assessment in the prediction of outcomes in MM.

Measurable Residual Disease (MRD) dynamics in Multiple Myeloma and the influence of Clonal Diversity Analyzed by Artificial Intelligence

Minimal residual disease (MRD) testing is a promising approach to tailor the treatment of multiple myeloma (MM). However, several major concerns remain to be addressed before moving it into daily practice, most of which stem from the dynamic natures of the MRD status. Thus, it is crucial to understand the MRD dynamics and propose its clinical implications. 

Dynamics of minimal residual disease and its clinical implications in multiple myeloma: A retrospective real-life analysis

Measurable residual disease (MRD) dynamics in multiple myeloma and the influence of clonal diversity analyzed by artificial intelligence

To the Editor: With the introduction of novel agents, more than 80% of patients with newly-diagnosed MM (NDMM) can achieve complete remission (CR) or better after first-line treatment. However, the majority of these patients eventually experience relapse, therefore calling for an additional approach to more precisely evaluate the depth of response and estimate the outcomes. Minimal residual disease (MRD) has been emerging as a reliable surrogate endpoint for progression-free survival (PFS) in clinical trials, while undetectable MRD strongly correlates with favorable outcomes, including both PFS and overall survival (OS). Over 50 ongoing phase III trials have included MRD as an endpoint to evaluate the depth of therapeutic response or assign MRD-directed treatment. Thus, MRD has been included into the revised International MyelomaWorking Group (IMWG) criteria for assessment of therapeutic response. Moreover, the prognostic value of MRD seems independent of baseline risk factors such as advanced disease stage (e.g., ISS or R-ISS III) and high-risk cytogenetic abnormality (HRCA). However, several key issues remain to be addressed for incorporating the MRD test into daily practice. To this end, we conducted this retrospective study to explore the significance of dynamic monitoring of the MRD status in the management of NDMM patients. This study was approved by the Institutional Review Board (IRB) of the First Hospital of Jilin University (approval # 2018-087). The study included patients with NDMM who had flow cytometry-based MRD test at least twice after initiation of treatment, between March 2015 and June 2019. CAs were assessed by FISH in CD138 cells isolated from bone marrow (BM) samples via routine diagnostic procedures. Among them, del(17p), t(4;14), t(14;16), and + 1q (1q21 gain/ amplification) were considered as HRCAs. Therapeutic responses were evaluated according to the IMWG consensus criteria. The first MRD test was conducted at the time when an objective response (e.g., PR, VGPR, CR, or sCR) was achieved during induction or consolidation. MRD was evaluated in bone marrow (BM) samples (1.5 10 cells per sample) using a modified 2-tube 8-color assay (tube 1: cKappa-FITC/cLambda-PE/CD138-Percp-cy5.5/CD27-PE-CY7/CD22APC/CD19-APC-H7/CD38-BV421/CD45-Viogreen; tube 2: CD56FITC/CD200-PE/CD28-Percp-cy5.5/CD117-PE-CY7/CD19-APC/ CD20-APC-H7/CD38-BV421/CD45-Viogreen.) by Navios flow cytometer (Beckman Coulter) in a single clinical laboratory. Two million cellular events were analyzed using the Kaluza Analysis Software to determine the percentage of phenotypically aberrant clonal plasma cells (PCs) in total nucleated cells. The limit of detection (LOD) achieved was calculated in each sample according to the following formula: (20/number of viable nucleated cells) 100. Undetectable MRD was defined as the absence of phenotypically aberrant clonal PCs or their percentage below the LOD (median 3.46 10 ), while detectable MRD was defined as the percentage of phenotypically aberrant clonal PCs equal to or greater than the LOD. Duration of undetectable MRD was defined as the time from the first detection of undetectable MRD to reappearance of detectable MRD. PFS was defined as the time from the first day of treatment until the date when disease progression, relapse, or death was first reported. OS was defined as the time from the first day of treatment until the date of death due to any cause or last follow-up. All statistical analyses were conducted using SPSS software (version 22.0) and R packages survival and survminer in R/Bioconductor (version 3.6.1). p < .05 was considered significant. A total of 220 patients with NDMM were assessed for MRD at least twice, with median of three times (range 2–11). The first test was conducted at initial objective responses, including PR (38.2%), very good partial response (VGPR, 19.1%), and ≥ CR (42.7%). Among them, 114 patients attained undetectable MRD (MRD ), while 106 patients had persistent MRD (MRD). To examine whether baseline variables would influence the MRD status, clinical characteristics and treatment were compared between MRD and MRD (Table S1). Interestingly, patients with ≥2 HRCAs (named double-hit) had lower rate of MRD attainment (25.0%) than those with one HRCA (53.9%; p = .007). With the median follow-up time of 33.3 months, the Kaplan– Meier curve revealed that attainment of MRD substantially prolonged PFS (Hazard ratio/HR, 0.19, 95% CI, 0.12–0.28, p < .0001; Figure S1A) and OS (HR, 0.25, 95% CI, 0.15–0.42, p < .0001; Figure S1B). Similar results were obtained in the subgroup of elderly patients (age ≥ 65 years), including PFS (p < .0001; Figure S1C) and OS (p = .0049; Figure S1D). Whereas the time-to-MRD varied, virtually all MRD patients achieved MRD within a year (70% within half a year) after initiation of treatment (not shown). Accordingly, 1-year landmark analysis was conducted and revealed that MRD were significantly associated with prolonged PFS (p < .001 for both time periods; not shown) and OS (Figure 1A) within or beyond 1 year. Three-year PFS rate of MRD and MRD patients was 58.9% versus 12.4%; 5-year OS rate of MRD and MRD patients was 67.5% versus 21.7%. Subgroup analysis according to the MRD status revealed that the favorable impact of MRD on both PFS (Figure S1E) and OS (Figure S1F) was consistent across all subgroups categorized by various risk factors at diagnosis. Following the univariate analysis Received: 14 November 2022 Revised: 5 December 2022 Accepted: 6 December 2022

Dynamic monitoring of minimal residual disease in newly‐diagnosed multiple myeloma

In recent years, molecular graph representation learning (GRL) has drawn much more attention in molecular property prediction (MPP) problems. The existing graph methods have demonstrated that 3D geometric information is significant for better performance in MPP. However, accurate 3D structures are often costly and time-consuming to obtain, limiting the large-scale application of GRL. It is an intuitive solution to train with 3D to 2D knowledge distillation and predict with only 2D inputs. But some challenging problems remain open for 3D to 2D distillation. One is that the 3D view is quite distinct from the 2D view, and the other is that the gradient magnitudes of atoms in distillation are discrepant and unstable due to the variable molecular size. To address these challenging problems, we exclusively propose a distillation framework that contains global molecular distillation and local atom distillation. We also provide a theoretical insight to justify how to coordinate atom and molecular information, which tackles the drawback of variable molecular size for atom information distillation. Experimental results on two popular molecular datasets demonstrate that our proposed model achieves superior performance over other methods. Specifically, on the largest MPP dataset PCQM4Mv2 served as an"ImageNet Large Scale Visual Recognition Challenge"in the field of graph ML, the proposed method achieved a 6.9% improvement compared with the best works. And we obtained fourth place with the MAE of 0.0734 on the test-challenge set for OGB-LSC 2022 Graph Regression Task. We will release the code soon.

Coordinating Cross-modal Distillation for Molecular Property Prediction

Handle everyday research tasks with reliable, citation-backed results

Your personal Research Agent to handle research tasks with citation-backed results

Popular Tasks used by Researchers

How can I help with your research?

Meet SciSpace

Get more enhanced response by uploading the PDFs you want me to reference.

No relevant PDFs in your library

SciSpace is the AI research assistant for academics. Run systematic literature reviews on 280M+ papers, and write papers with cited sources. Trusted by 1M+ students, PhDs & researchers.

SciSpace | AI for Research

Analyze PDFs

Code & Manuscripts

Funding & Grants

Literature & Patents

Medical & Clinical Data

Systematic Review

Visualize & Present

Web & Data

Build a Google Scholar-like website for your research.

Build a website

Create charts and images for your research

Create a Chart

Write a paper for submission to a journal

Draft a manuscript

Patent Search

Design eye-catching scientific posters in minutes.

Scientific Poster Generation

Systematic Literature Review

One task is running at the moment. Your messages will be shown right after.

Drag and drop or click here to browse

Loved by <highlight>1 million+</highlight> researchers

Extract a list of specific topics and their sources from unstructured text

Topics

Compare and analyze relevant papers that matches with your search

Papers

Get insights from PDFs and bookmarked papers from your library

My library

Recent searches

Try searching for:

Catch AI-generated content in scholarly and non-scholarly content

{ai} Detector

Ai Writer

Get PDF Summaries, highlighted text explanations 

Chat with PDF

Effortlessly create in-text citations and bibliographies in APA and 2,500 other formats

Citation generator

Get explanations, summaries, and answers on academic papers

Ease up your research workflow with {scispace}'s cohort of exciting AI tools

Elevate your academic writing skills and convey your ideas the way you want

Paraphraser

Explore our range of reading and writing tools

Your file is being prepared and should be ready in a few minutes. If it's a large file, it might take a bit longer. You can close this window, and we'll email you the file when it's done.

You have reached a maximum limit of {limit} columns in the table. Remove at least 1 column to add or create another one.

Meng Liu

Author Tools

Chat about Author

Papers

Dynamic monitoring of minimal residual disease in newly‐diagnosed multiple myeloma

Coordinating Cross-modal Distillation for Molecular Property Prediction