scholarly journals Digital medicine and the curse of dimensionality

2021 ◽  
Vol 4 (1) ◽  
Author(s):  
Visar Berisha ◽  
Chelsea Krantsevich ◽  
P. Richard Hahn ◽  
Shira Hahn ◽  
Gautam Dasarathy ◽  
...  
Keyword(s):  
Real World ◽  
Digital Health ◽  
Model Performance ◽  
High Volume ◽  
Curse Of Dimensionality ◽  
Health Data ◽  
Limiting Factors ◽  
High Dimensional ◽  
Health State ◽  
Out Of Sample

AbstractDigital health data are multimodal and high-dimensional. A patient’s health state can be characterized by a multitude of signals including medical imaging, clinical variables, genome sequencing, conversations between clinicians and patients, and continuous signals from wearables, among others. This high volume, personalized data stream aggregated over patients’ lives has spurred interest in developing new artificial intelligence (AI) models for higher-precision diagnosis, prognosis, and tracking. While the promise of these algorithms is undeniable, their dissemination and adoption have been slow, owing partially to unpredictable AI model performance once deployed in the real world. We posit that one of the rate-limiting factors in developing algorithms that generalize to real-world scenarios is the very attribute that makes the data exciting—their high-dimensional nature. This paper considers how the large number of features in vast digital health data can challenge the development of robust AI models—a phenomenon known as “the curse of dimensionality” in statistical learning theory. We provide an overview of the curse of dimensionality in the context of digital health, demonstrate how it can negatively impact out-of-sample performance, and highlight important considerations for researchers and algorithm designers.

scholarly journals Overcoming the Curse of Dimensionality in the Numerical Approximation of Parabolic Partial Differential Equations with Gradient-Dependent Nonlinearities

2021 ◽  
Author(s):  
Martin Hutzenthaler ◽  
Arnulf Jentzen ◽  
Thomas Kruse
Keyword(s):  
Partial Differential Equations ◽  
Approximation Algorithm ◽  
Real World ◽  
Curse Of Dimensionality ◽  
Nonlinear Pdes ◽  
High Dimensional ◽  
Heat Equations ◽  
Time Horizons ◽  
Nonlinear Heat Equations ◽  
General Time

AbstractPartial differential equations (PDEs) are a fundamental tool in the modeling of many real-world phenomena. In a number of such real-world phenomena the PDEs under consideration contain gradient-dependent nonlinearities and are high-dimensional. Such high-dimensional nonlinear PDEs can in nearly all cases not be solved explicitly, and it is one of the most challenging tasks in applied mathematics to solve high-dimensional nonlinear PDEs approximately. It is especially very challenging to design approximation algorithms for nonlinear PDEs for which one can rigorously prove that they do overcome the so-called curse of dimensionality in the sense that the number of computational operations of the approximation algorithm needed to achieve an approximation precision of size $${\varepsilon }> 0$$ ε > 0 grows at most polynomially in both the PDE dimension $$d \in \mathbb {N}$$ d ∈ N and the reciprocal of the prescribed approximation accuracy $${\varepsilon }$$ ε . In particular, to the best of our knowledge there exists no approximation algorithm in the scientific literature which has been proven to overcome the curse of dimensionality in the case of a class of nonlinear PDEs with general time horizons and gradient-dependent nonlinearities. It is the key contribution of this article to overcome this difficulty. More specifically, it is the key contribution of this article (i) to propose a new full-history recursive multilevel Picard approximation algorithm for high-dimensional nonlinear heat equations with general time horizons and gradient-dependent nonlinearities and (ii) to rigorously prove that this full-history recursive multilevel Picard approximation algorithm does indeed overcome the curse of dimensionality in the case of such nonlinear heat equations with gradient-dependent nonlinearities.

scholarly journals Ethical Considerations When Creating Evidence from Real World Digital Health Data

2020 ◽  
Author(s):  
C. Nebeker ◽  
Victoria Leavy ◽  
Eva Roitmann ◽  
Steven Steinhubl
Keyword(s):  
Data Sharing ◽  
Real World ◽  
Digital Health ◽  
Health Data ◽  
Personal Health ◽  
Real World Evidence ◽  
Self Study ◽  
Health Related Research ◽  
Health Related ◽  
English Speaking

AbstractBackgroundPersonal health data (PHD) are collected using digital self-tracking technologies and present opportunities to increase self-knowledge and, also biometric surveillance. PHD become “big” data and are used in health-related research studies. We surveyed consumers regarding expectations regarding consent and sharing of PHD for biomedical research.MethodsData sharing preferences were assessed via an 11-item survey. The survey link was emailed to 89539 English-speaking Withings product users. Responses were accepted for 5 weeks.Descriptive statistics were calculated using Excel and qualitative data were analyzed to provide additional context.ResultsNearly 1640 people or 5.7% of invitees responded representing 62 countries with 80% identifying as Caucasian, 75% male with 78% being college educated. The majority were agreeable to having their data shared with researchers to advance knowledge and improve health care.Participants responding to open ended items (N=247) appeared unaware that the company had access to their personal health data.ConclusionsWhile the majority of respondents were in favor of data sharing, individuals expressed concerns about the ability to de-identify data and associated risks of re-identification as well as an interest in having some control over the use of “their” data. Given consumer misconception about data ownership, access and use, efforts to increase transparency when interacting with individual digital health data must be prioritized. Moreover, the basic ethical principle of “respect for persons” demonstrated via the informed consent process will be critical in advancing the adoption of digital technologies that create real-world evidence and advance opportunities for N-of-1 self-study.

scholarly journals Predicting Future Occurrence of Acute Hypotensive Episodes Using Noninvasive and Invasive Features

2021 ◽  
Vol 186 (Supplement_1) ◽  
pp. 445-451
Author(s):  
Yifei Sun ◽  
Navid Rashedi ◽  
Vikrant Vaze ◽  
Parikshit Shah ◽  
Ryan Halter ◽  
...  
Keyword(s):  
Machine Learning ◽  
Support Vector Machine ◽  
Real World ◽  
Short Term Memory ◽  
Model Performance ◽  
Learning Technologies ◽  
Machine Learning Algorithms ◽  
Support Vector ◽  
K Nearest Neighbor ◽  
Continuous Map

ABSTRACT Introduction Early prediction of the acute hypotensive episode (AHE) in critically ill patients has the potential to improve outcomes. In this study, we apply different machine learning algorithms to the MIMIC III Physionet dataset, containing more than 60,000 real-world intensive care unit records, to test commonly used machine learning technologies and compare their performances. Materials and Methods Five classification methods including K-nearest neighbor, logistic regression, support vector machine, random forest, and a deep learning method called long short-term memory are applied to predict an AHE 30 minutes in advance. An analysis comparing model performance when including versus excluding invasive features was conducted. To further study the pattern of the underlying mean arterial pressure (MAP), we apply a regression method to predict the continuous MAP values using linear regression over the next 60 minutes. Results Support vector machine yields the best performance in terms of recall (84%). Including the invasive features in the classification improves the performance significantly with both recall and precision increasing by more than 20 percentage points. We were able to predict the MAP with a root mean square error (a frequently used measure of the differences between the predicted values and the observed values) of 10 mmHg 60 minutes in the future. After converting continuous MAP predictions into AHE binary predictions, we achieve a 91% recall and 68% precision. In addition to predicting AHE, the MAP predictions provide clinically useful information regarding the timing and severity of the AHE occurrence. Conclusion We were able to predict AHE with precision and recall above 80% 30 minutes in advance with the large real-world dataset. The prediction of regression model can provide a more fine-grained, interpretable signal to practitioners. Model performance is improved by the inclusion of invasive features in predicting AHE, when compared to predicting the AHE based on only the available, restricted set of noninvasive technologies. This demonstrates the importance of exploring more noninvasive technologies for AHE prediction.

scholarly journals SAT0432 EFFECT OF SEX ON DISEASE CHARACTERISTICS AND DISEASE IMPACT IN PATIENTS WITH PSORIATIC ARTHRITIS (PsA): INSIGHTS FROM THE REAL-WORLD, OBSERVATIONAL MULTINATIONAL PsABio COHORT

2020 ◽  
Vol 79 (Suppl 1) ◽  
pp. 1171.1-1173
Author(s):  
M. T. Nurmohamed ◽  
I. Van der Horst-Bruinsma ◽  
A. W. Van Kuijk ◽  
S. Siebert ◽  
P. Bergmans ◽  
...  
Keyword(s):  
Psoriatic Arthritis ◽  
Disease Activity ◽  
Physical Functioning ◽  
Real World ◽  
Comprehensive Treatment ◽  
Diagnosis Delay ◽  
Health State ◽  
Research Support ◽  
Disease Characteristics ◽  
Disease Impact

Background:Female sex has been associated with more severe disease and poorer treatment outcomes in PsA. These observations are often based on small populations or national cohorts/registries.Objectives:To investigate the effects of sex on disease characteristics and disease impact in PsA, using data of 929 consecutive patients (pts) from PsABio.Methods:PsABio is a real-world, non-interventional European study in PsA pts treated with UST or TNFi based on their rheumatologist’s choice. Observed male and female baseline (BL) data were described and compared using 95% CI.Results:Women in PsABio (n=512 [55%]) were numerically older than men (mean [SD]: 50.5 [12.7] / 48.7 [12.3] years, respectively). Women were more obese (BMI >30), % (95% CI): F: 35 (30, 39), M: 24 (20, 29), men more overweight (BMI >25–30): F: 31 (27, 36), M:51 (46, 57). Age at diagnosis, delay from first symptom to diagnosis, and disease duration were similar for both sexes.Women entered PsABio more often on 3rd line treatment, whereas men started on 1st-line biologic treatment more often (F/M 1st line 47%/55%; 2nd line 34%/33%; 3rd line 20%/12%). Numerically, concomitant MTX was given more often to women vs men (32% vs 27%). At BL, 60% of women and 64% of men were on NSAIDs; 7.9% and 2.5% on antidepressant drugs. Women had significantly more comorbidities, with numerically more cardiovascular disease and anxiety/depression, and 3 times more IBD.Women had significantly higher 68 tender joint counts (TJC): 13.0 vs 10.4, while 66 swollen joint counts were not significantly different: 5.8 vs 5.5. Axial or combined axial-peripheral disease was similarly frequent, in 29% of women and 26% of men (Figs. 1, 2).Clinical Disease Activity index for PSoriatic Arthritis (cDAPSA) was higher in women (31.8 vs 27.3); pt-reported levels of pain, global disease activity (VAS scales) and higher TJC contributed to this. While enthesitis prevalence (based on Leeds Enthesitis Index) was comparable, men had significantly more frequent dactylitis, nail disease and worse skin psoriasis. At BL, 3.4% of women vs 7.1% of men, were in MDA.Regarding physical functioning (HAQ-DI), impact of disease (PSAID-12) and quality of life (EQ5D-3L health state), women with PsA starting a biologic (b)DMARD, expressed significantly greater negative impact and more limitations due to their disease (Fig. 2).Conclusion:In routine care, women with PsA starting a bDMARD presented with worse outcomes over a range of assessments compared with men (higher pt-reported pain and disease activity, TJC, and worse physical functioning and QoL), while men had worse dactylitis and psoriasis. Follow-up analysis will report whether the effects of biologic therapy are different in both sexes. The increased prevalence of associated features related to pain and impact on functioning and QoL may indicate the need for a more comprehensive treatment approach for women to avoid unnecessary and premature bDMARD stop or switch.Acknowledgments:This study was funded by Janssen.Disclosure of Interests:Michael T Nurmohamed Grant/research support from: Abbvie, Bristol-Myers Squibb, Celltrion, GlaxoSmithKline, Jansen, Eli Lilly, Menarini, Merck Sharp & Dohme, Mundipharma, Pfizer, Roche, Sanofi, USB, Consultant of: Abbvie, Bristol-Myers Squibb, Celltrion, GlaxoSmithKline, Jansen, Eli Lilly, Menarini, Merck Sharp & Dohme, Mundipharma, Pfizer, Roche, Sanofi, USB, Speakers bureau: Abbvie, Bristol-Myers Squibb, Celltrion, GlaxoSmithKline, Jansen, Eli Lilly, Menarini, Merck Sharp & Dohme, Mundipharma, Pfizer, Roche, Sanofi, USB, Irene van der Horst-Bruinsma Grant/research support from: AbbVie, Novartis, Eli Lilly, Bristol-Myers Squibb, MSD, Pfizer, UCB Pharma, Consultant of: AbbVie, Novartis, Eli Lilly, Bristol-Myers Squibb, MSD, Pfizer, UCB Pharma, Arno WR van Kuijk Grant/research support from: Janssen, Stefan Siebert Grant/research support from: BMS, Boehringer Ingelheim, Celgene, GlaxoSmithKline, Janssen, Novartis, Pfizer, UCB, Consultant of: AbbVie, Boehringer Ingelheim, Janssen, Novartis, Pfizer, UCB, Speakers bureau: AbbVie, Celgene, Janssen, Novartis, Paul Bergmans Shareholder of: Johnson & Johnson, Employee of: Janssen, Kurt de Vlam Consultant of: Celgene Corporation, Eli Lilly, Novartis, Pfizer, UCB – consultant, Speakers bureau: Celgene Corporation, Eli Lilly, Novartis, Pfizer, UCB – speakers bureau and honoraria, Elisa Gremese Consultant of: AbbVie, Bristol-Myers Squibb, Celgene, Eli Lilly, Janssen, Merck Sharp & Dohme, Novartis, Sanofi, UCB, Roche, Pfizer, Speakers bureau: AbbVie, Bristol-Myers Squibb, Celgene, Eli Lilly, Janssen, Merck Sharp & Dohme, Novartis, Sanofi, UCB, Roche, Pfizer, Beatriz Joven-Ibáñez Speakers bureau: Abbvie, Celgene, Janssen, Merck Sharp & Dohme, Novartis, Pfizer, Tatiana Korotaeva Grant/research support from: Pfizer, Consultant of: Abbvie, BIOCAD, Bristol-Myers Squibb, Celgene, Eli Lilly, Janssen, Merck Sharp & Dohme, Novartis, Novartis-Sandoz, Pfizer, UCB, Speakers bureau: Abbvie, BIOCAD, Bristol-Myers Squibb, Celgene, Eli Lilly, Janssen, Merck Sharp & Dohme, Novartis, Novartis-Sandoz, Pfizer, UCB, Wim Noel Employee of: Janssen Pharmaceuticals NV, Petros Sfikakis Grant/research support from: Grant/research support from Abvie, Novartis, MSD, Actelion, Amgen, Pfizer, Janssen Pharmaceutical, UCB, Elke Theander Employee of: Janssen-Cilag Sweden AB, Josef S. Smolen Grant/research support from: AbbVie, AstraZeneca, Celgene, Celltrion, Chugai, Eli Lilly, Gilead, ILTOO, Janssen, Novartis-Sandoz, Pfizer Inc, Samsung, Sanofi, Consultant of: AbbVie, AstraZeneca, Celgene, Celltrion, Chugai, Eli Lilly, Gilead, ILTOO, Janssen, Novartis-Sandoz, Pfizer Inc, Samsung, Sanofi, Laure Gossec Grant/research support from: Lilly, Mylan, Pfizer, Sandoz, Consultant of: AbbVie, Amgen, Biogen, Celgene, Janssen, Lilly, Novartis, Pfizer, Sandoz, Sanofi-Aventis, UCB

scholarly journals Assessing and piloting interoperability for population health research

2020 ◽  
Vol 30 (Supplement_5) ◽  
Author(s):  
F Estupiñán-Romero ◽  
J Gonzalez-García ◽  
E Bernal-Delgado
Keyword(s):  
Case Studies ◽  
Population Health ◽  
Health Research ◽  
Network Architecture ◽  
Digital Health ◽  
Data Extraction ◽  
Health Data ◽  
Research Network ◽  
Technical Solution ◽  
Privacy And Security

Abstract Issue/problem Interoperability is paramount when reusing health data from multiple data sources and becomes vital when the scope is cross-national. We aimed at piloting interoperability solutions building on three case studies relevant to population health research. Interoperability lies on four pillars; so: a) Legal frame (i.e., compliance with the GDPR, privacy- and security-by-design, and ethical standards); b) Organizational structure (e.g., availability and access to digital health data and governance of health information systems); c) Semantic developments (e.g., existence of metadata, availability of standards, data quality issues, coherence between data models and research purposes); and, d) Technical environment (e.g., how well documented are data processes, which are the dependencies linked to software components or alignment to standards). Results We have developed a federated research network architecture with 10 hubs each from a different country. This architecture has implied: a) the design of the data model that address the research questions; b) developing, distributing and deploying scripts for data extraction, transformation and analysis; and, c) retrieving the shared results for comparison or pooled meta-analysis. Lessons The development of a federated architecture for population health research is a technical solution that allows full compliance with interoperability pillars. The deployment of this type of solution where data remain in house under the governance and legal requirements of the data owners, and scripts for data extraction and analysis are shared across hubs, requires the implementation of capacity building measures. Key messages Population health research will benefit from the development of federated architectures that provide solutions to interoperability challenges. Case studies conducted within InfAct are providing valuable lessons to advance the design of a future pan-European research infrastructure.

Regulations and Norms for Reuse of Residual Clinical Biospecimens and Health Data

2021 ◽  
pp. 019394592110292
Author(s):  
Elizabeth E. Umberfield ◽  
Sharon L. R. Kardia ◽  
Yun Jiang ◽  
Andrea K. Thomer ◽  
Marcelline R. Harris
Keyword(s):  
Informed Consent ◽  
Scoping Review ◽  
Real World ◽  
Genetic Information ◽  
Final Analysis ◽  
Health Data ◽  
Federal Regulations ◽  
Secondary Research ◽  
Domain Experts ◽  
Key Issues

Nurse scientists are increasingly interested in conducting secondary research using real world collections of biospecimens and health data. The purposes of this scoping review are to (a) identify federal regulations and norms that bear authority or give guidance over reuse of residual clinical biospecimens and health data, (b) summarize domain experts’ interpretations of permissions of such reuse, and (c) summarize key issues for interpreting regulations and norms. Final analysis included 25 manuscripts and 23 regulations and norms. This review illustrates contextual complexity for reusing residual clinical biospecimens and health data, and explores issues such as privacy, confidentiality, and deriving genetic information from biospecimens. Inconsistencies make it difficult to interpret, which regulations or norms apply, or if applicable regulations or norms are congruent. Tools are necessary to support consistent, expert-informed consent processes and downstream reuse of residual clinical biospecimens and health data by nurse scientists.

scholarly journals Characterizing COVID-19 and Influenza Illnesses in the Real World via Person-Generated Health Data

Patterns ◽  
2020 ◽  
pp. 100188
Author(s):  
Allison Shapiro ◽  
Nicole Marinsek ◽  
Ieuan Clay ◽  
Benjamin Bradshaw ◽  
Ernesto Ramirez ◽  
...  
Keyword(s):  
Real World ◽  
Health Data ◽  
The Real

scholarly journals Can pragmatic research, real-world data and digital technologies aid the development of psychedelic medicine?

2021 ◽  
pp. 026988112110085
Author(s):  
Robin L Carhart-Harris ◽  
Anne C Wagner ◽  
Manish Agrawal ◽  
Hannes Kettner ◽  
Jerold F Rosenbaum ◽  
...  
Keyword(s):  
Real World ◽  
Best Practice ◽  
Digital Health ◽  
Real World Data ◽  
Treatment Protocols ◽  
World Data ◽  
Cost Efficient ◽  
Personalised Treatment ◽  
New Research ◽  
Confirmatory Trials

Favourable regulatory assessments, liberal policy changes, new research centres and substantial commercial investment signal that psychedelic therapy is making a major comeback. Positive findings from modern trials are catalysing developments, but it is questionable whether current confirmatory trials are sufficient for advancing our understanding of safety and best practice. Here we suggest supplementing traditional confirmatory trials with pragmatic trials, real-world data initiatives and digital health solutions to better support the discovery of optimal and personalised treatment protocols and parameters. These recommendations are intended to help support the development of safe, effective and cost-efficient psychedelic therapy, which, given its history, is vulnerable to excesses of hype and regulation.

scholarly journals Visual Analysis of Biomarkers Reveals Differences in Lipid Profiles and Liver Enzymes before and after Gastric Sleeve and Bypass

Obesity Facts ◽  
2021 ◽  
pp. 1-11
Author(s):  
Marijn Marthe Georgine van Berckel ◽  
Saskia L.M. van Loon ◽  
Arjen-Kars Boer ◽  
Volkher Scharnhorst ◽  
Simon W. Nienhuijs
Keyword(s):  
Bariatric Surgery ◽  
Visual Analytics ◽  
Biochemical Markers ◽  
Visual Analysis ◽  
Large Population ◽  
High Volume ◽  
Cholesterol Concentration ◽  
Health State ◽  
Data Set ◽  
Before And After

<b><i>Introduction:</i></b> Bariatric surgery results in both intentional and unintentional metabolic changes. In a high-volume bariatric center, extensive laboratory panels are used to monitor these changes pre- and postoperatively. Consecutive measurements of relevant biochemical markers allow exploration of the health state of bariatric patients and comparison of different patient groups. <b><i>Objective:</i></b> The objective of this study is to compare biomarker distributions over time between 2 common bariatric procedures, i.e., sleeve gastrectomy (SG) and gastric bypass (RYGB), using visual analytics. <b><i>Methods:</i></b> Both pre- and postsurgical (6, 12, and 24 months) data of all patients who underwent primary bariatric surgery were collected retrospectively. The distribution and evolution of different biochemical markers were compared before and after surgery using asymmetric beanplots in order to evaluate the effect of primary SG and RYGB. A beanplot is an alternative to the boxplot that allows an easy and thorough visual comparison of univariate data. <b><i>Results:</i></b> In total, 1,237 patients (659 SG and 578 RYGB) were included. The sleeve and bypass groups were comparable in terms of age and the prevalence of comorbidities. The mean presurgical BMI and the percentage of males were higher in the sleeve group. The effect of surgery on lowering of glycated hemoglobin was similar for both surgery types. After RYGB surgery, the decrease in the cholesterol concentration was larger than after SG. The enzymatic activity of aspartate aminotransferase, alanine aminotransferase, and alkaline phosphate in sleeve patients was higher presurgically but lower postsurgically compared to bypass values. <b><i>Conclusions:</i></b> Beanplots allow intuitive visualization of population distributions. Analysis of this large population-based data set using beanplots suggests comparable efficacies of both types of surgery in reducing diabetes. RYGB surgery reduced dyslipidemia more effectively than SG. The trend toward a larger decrease in liver enzyme activities following SG is a subject for further investigation.

Simple Multivariate Conditional Covariance Dynamics Using Hyperbolically Weighted Moving Averages

2020 ◽  
Vol 0 (0) ◽  
Author(s):  
Hiroyuki Kawakatsu
Keyword(s):  
Moving Average ◽  
Curse Of Dimensionality ◽  
Moving Averages ◽  
Conditional Covariance ◽  
Arch Models ◽  
Empirical Application ◽  
Proposed Model ◽  
Out Of Sample ◽  
Weight Model ◽  
Weighted Moving Average

AbstractThis paper considers a class of multivariate ARCH models with scalar weights. A new specification with hyperbolic weighted moving average (HWMA) is proposed as an analogue of the EWMA model. Despite the restrictive dynamics of a scalar weight model, the proposed model has a number of advantages that can deal with the curse of dimensionality. The empirical application illustrates that the (pseudo) out-of-sample multistep forecasts can be surprisingly more accurate than those from the DCC model.

Sign in / Sign up

Export Citation Format

Share Document