PaperlayerView exact revision on arXiv

Source-linked AI summary

Going Digital: A Survey on Digitalization and Large Scale Data Analytics in Healthcare

Volker Tresp, J. Marc Overhage, Markus Bundschus, Shahrooz Rabizadeh, Peter A. Fasching, Shipeng Yu

arXiv:1606.08075v3cs.CY

TL;DR

Linking molecular data with clinical outcomes remains an important challenge for personalized healthcare. This paper surveys efforts to improve care through digitalization, data exchange, and analytics. It concludes that digitalization and large-scale analytics will be key technologies in healthcare’s coming changes, while their benefits need to be apparent to caregivers and patients.

  • Problem

    Precision healthcare requires linking molecular data with clinical outcomes, but questions remain about integrating genomic data into EHR records.

    Source: F. Digitization Challenges in… · p. 18

  • Method

    The paper surveys efforts to improve healthcare through more effective processes, standardized data formats, trusted data-exchange networks, and changes in policies and reimbursement rules.

    Source: §VIII · p. 21

  • Results

    Digitalization and large-scale data analytics will be among the key technologies in healthcare’s coming changes.

    Source: §VIII · p. 20

  • Takeaways & Limitations

    Caregivers and patients need to experience readily apparent benefits, rather than only additional administrative burdens, for the vision of better and more efficient healthcare to gain support.

    Source: §VIII · p. 21

Abstract

from arXiv · show

We provide an overview of the recent trends towards digitalization and large scale data analytics in healthcare. It is expected that these trends are instrumental in the dramatic changes in the way healthcare will be organized in the future. We discuss the recent political initiatives designed to shift care delivery processes from paper to electronic, with the goals of more effective treatments with better outcomes; cost pressure is a major driver of innovation. We describe newly developed networks of healthcare providers, research organizations and commercial vendors to jointly analyze data for the development of decision support systems. We address the trend towards continuous healthcare where health is monitored by wearable and stationary devices; a related development is that patients increasingly assume responsibility for their own health data. Finally we discuss recent initiatives towards a personalized medicine, based on advances in molecular medicine, data management, and data analytics.

I. INTRODUCTION

Healthcare is moving toward abundant digital data and large-scale analytics, driven by policy initiatives and cost pressure. The paper surveys how these developments may reshape care, from clinical records and data exchange to patient-managed data, continuous monitoring, and personalized medicine.

I. INTRODUCTION

  • Digital health data are becoming abundant and high-dimensional, shifting the challenge from study design and acquisition toward data volume and management.The paper describes the resulting management and analysis of digital health data as Healthcare Big Data,,.Source: §I · p. 1
  • Policy initiatives aim to move care from paper to electronic records for more effective treatment and better outcomes, while cost pressure drives innovation.The HITECH Act emphasizes meaningful use of interoperable EHRs to exchange information across institutions; privacy, ownership, and security issues remain to be resolved.Source: §I · p. 1
  • Healthcare Big Data relies on varied, detailed patient information and large datasets for analytics, while key challenges include acquiring, managing, and communicating recommendations from those data.Examples of recommendations include lifestyle guidance, reminders, screening, referrals, and treatment recommendations.Source: §I · p. 1
  • The paper presents digital health as a potential source of dramatic future changes in healthcare and organizes its survey around six digitalization sources.Source: §I · p. 1
  • The overview covers clinical data management, provider-researcher-vendor collaborations, cross-organization exchange, and patient-centered platforms for managing and sharing health data.It also discusses biases, hidden confounders, and batch effects when integrating observations from different sources.Source: §I · pp. 1–2
  • Everyday mobile and ambient sensing supports continuous health monitoring and prevention, while molecular profiles and large-scale analytics contribute to personalized care.The paper notes that personalization increases decision complexity and that it focuses primarily on the U.S., while highlighting relevant developments elsewhere.Source: §I · p. 2

II. DIGITIZING HEALTHCARE DATA

Digitized healthcare data can support learning systems, analytics, and clinical decision support, but their value depends on usable records and structured, interoperable data. The section surveys the EHR and the challenges of capturing clinical information in computable form.

A. Motivation

  • A learning healthcare system continuously analyzes data from care to inform how care delivery should evolve, while analytics can support clinical and research decisions.Descriptive, predictive, and prescriptive analytics serve different purposes; prediction across clinics may be unreliable when patient profiles differ, and prescriptive analysis requires attention to hidden confounders.Source: A. Motivation · pp. 2–3
  • 40–50% reduction in readmissions was reported for congestive-heart-failure patients identified by insurers’ analytics.Source: A. Motivation · p. 3

B. The Electronic Health Record

  • Advanced EHRs make machine-readable structured data and semantically annotated reports accessible to analytic tools, whereas scanned paper documents offer little support for analysis.U.S. EHR adoption has broadened, but physician enthusiasm has not improved, with usability, functionality, interoperability, leakage, and compliance concerns noted.Source: B. The Electronic Health Record · p. 3

C. Structured Data Capture

  • Provider entry of structured data takes time, and detailed clinical narratives contain sequential observations and subtle language that current formal representations struggle to express.Contemporary EHRs are estimated to require an additional 48 minutes per day, much of it for documentation; text extraction advances have been incremental, and structured reporting is not standard partly because it takes providers longer.Source: C. Structured Data Capture · pp. 3–4
  • Clinical data use requires mapping among multiple coding systems, while images, text, and signals offer additional sources from which structured data may be derived.Radiology and pathology image analysis can generate qualitative and quantitative labels for integrated analytics studies.Source: C. Structured Data Capture · pp. 3–4

D. Data Silos

Fragmented clinical data are difficult to combine across departmental systems, research databases, and privacy constraints. Integrated care networks can broaden and enrich the data available for analytics, but data quality and integration remain important boundaries.

D. Data Silos

  • Departmental and organizational data silos fragment patient information, and integrating them is difficult because systems may code data differently.Organizations have sometimes accepted lesser functionality to use integrated EHR platforms with a common database across departments.Source: D. Data Silos · p. 4

E. Clinical Data Integration Efforts

  • Research databases can support analytics but may be demanding to build across heterogeneous departmental sources; billing data are structured yet may omit clinical detail or reflect reimbursement priorities, and event timing can be hard to reconstruct.Source: E. Clinical Data Integration Efforts · p. 5

F. Privacy Protection and De-identification

  • De-identification can reduce data usefulness, and it is particularly difficult for clinical text and genomic data because identifiers may appear in text and genetic profiles are unique.Source: F. Privacy Protection and… · p. 5

III. MOBILIZING DATA IN A TRUSTED NETWORK

  • Integrated care networks enable evaluation across multiple sources and larger patient samples, while longer records within a system may make individual patient data more complete.The paper describes projects pairing clinic networks with research centers that contribute analytics, machine-learning, and medical-informatics expertise.Source: §III · p. 5

A. The Pittsburgh Health Data Alliance

Healthcare Big Data collaborations bring providers, researchers, and commercial partners together to analyze diverse clinical data and develop decision support and personalized care. Projects range from shared data infrastructure to clinical applications, while documentation quality remains a constraint.

A. The Pittsburgh Health Data Alliance

  • The Pittsburgh Health Data Alliance combines CMU, Pitt, and UPMC to analyze diverse healthcare data for evidence-based, patient-centered, and personalized care.Its intended outputs also include spin-off companies and regional economic development.Source: A. The Pittsburgh Health Data Alliance · p. 5
  • The alliance attracted IT vendors supplying database, business-intelligence, and patient-record integration platforms as providers sought Big Data capabilities.The vendors were also exploring future healthcare business opportunities.Source: A. The Pittsburgh Health Data Alliance · p. 5

B. The Mayo Project

  • The Mayo project establishes a Big Data computing center and high-speed research network, including infrastructure for off-campus data access.Its major project is the Knowledge Engine for Genomics, or KnowEnG.Source: B. The Mayo Project · p. 6

C. Neonatal Intensive Care at Kaiser Permanente

  • Kaiser Permanente’s Newborn Sepsis Calculator estimates neonatal sepsis probability to help care teams decide which newborns to evaluate and treat.It responds to a practice in which a mother’s elevated temperature may signal increased neonatal sepsis risk.Source: C. Neonatal Intensive Care at Kaiser… · p. 6

D. Indiana Network for Patient Care

  • The Indianapolis Network for Patient Care (INPC), implemented in 1995 for point-of-care diagnosis and treatment, had over 100 participating entities in 2016 and stored more than 4.7 billion standardized records.Its repository included over 118 million text reports from almost 15 million unique patients.Source: D. Indiana Network for Patient Care · p. 6

E. Clinical Data Intelligence

  • Clinical Data Intelligence’s integrated learning and decision system analyzes structured EHR data, reports, guidelines, and molecular data for predictive and prescriptive functions.The project also emphasizes terminology, metadata extraction, guideline integration, and shared research infrastructure.Source: E. Clinical Data Intelligence · pp. 6–7
  • Clinical Data Intelligence aims to improve documentation requirements because symptoms and outcomes are not consistently recorded, complicating analytics.The project’s use cases include renal-transplant drug interactions and adverse reactions, as well as breast-cancer risk factors, therapy evaluation, and side-effect prediction.Source: E. Clinical Data Intelligence · p. 7

F. Related Initiatives and Projects

Healthcare initiatives use shared analytics to improve care and operational performance, while observational Big Data studies offer broad patient coverage but face confounding, bias, interpretability, and acceptance concerns. The examples illustrate both practical applications and limits on drawing causal conclusions.

F. Related Initiatives and Projects

  • Across four HVHC hospital sites, sharing the best-performing site’s knee-replacement practices helped all four reduce length of stay by one day.The initial comparison found different costs and processes among the sites.Source: F. Related Initiatives and Projects · p. 7
  • Penn Medicine’s model detected 80 percent of severe sepsis cases as much as 30 hours before septic shock, versus two hours with traditional identification methods.The model used six vital-sign measurements, lab values, and more than 200 clinical variables.Source: F. Related Initiatives and Projects · p. 7

G. Comments on the Value of Big Data Studies

  • RCTs can minimize hidden confounding through careful design, but often use small, relatively healthy and homogeneous samples, limiting interaction analysis and personalized findings.The passage also notes recruitment costs and delays before findings enter practice.Source: G. Comments on the Value of Big Data… · p. 7
  • Big Data analyses can draw on larger, more varied patient populations and support personalized conclusions, but hidden confounders can produce correlations without causal effects.Multivariate models may help account for variables used in clinical decisions, though symptoms and complaints are often undocumented.Source: G. Comments on the Value of Big Data… · pp. 7–8
  • Combining data across institutions can introduce batch effects, which the authors say careful statistical analysis can address.The paper notes that clinicians’ acceptance of Big Data evidence remains unclear and that hypothesis generation is generally accepted when followed by clinical validation.Source: G. Comments on the Value of Big Data… · p. 8
  • High-dimensional models may predict well while remaining difficult for clinicians to interpret, and treatment prioritization may challenge medical tradition.The model may infer latent causes from observed surrogates when diagnostic tests do not fully describe health status.Source: G. Comments on the Value of Big Data… · p. 8

IV. OUT WITH THE DATA

Data leaving clinical systems support payer analysis, registry research, and information exchange across organizations. These sources can inform care and policy, but their usefulness depends on data quality, safeguards, and appropriate interpretation.

A. Introduction

  • The section covers payer-accessible data, registries, and information reported to healthcare agencies, including their uses in analysis, quality control, and policy.It also addresses privacy and data safety.Source: A. Introduction · p. 8

B. Data Accessible to Payers: Billing Data

  • Fraud analytics detects known patterns, prioritizes suspicious cases, and searches for new schemes using clinical pathways, temporal charging patterns, and provider networks.Despite these approaches, only a few percent of occurring fraud is estimated to be detected, amid fragmentation and a substantial grey zone.Source: B. Data Accessible to Payers: Billing… · p. 9

C. Registries

  • Population-based cancer registries track incident cases to identify clusters, generate hypotheses, monitor therapies and prevention, and inform public-health decisions.Their conclusions depend on complete, high-quality data, and registry analyses must account for possible confounders.Source: C. Registries · p. 9

D. HIE

  • HIE mobilizes digital healthcare information across organizations through centralized or federated architectures.Centralized systems hold complete copies of records, while federated systems leave providers responsible for their records, availability, and common standards.Source: D. HIE · p. 9
  • A nationwide health information network is intended to give physicians access to patients’ complete histories while preserving privacy and supporting a learning healthcare system.Patient participation may use opt-in or opt-out consent models.Source: D. HIE · p. 9

E. Care.Data

Care.Data proposed integrating health and social-care information for research, planning, and service comparisons, but concerns over consent, data use, and re-identification accompanied the project’s reported management problems. Incentive programs promote electronic records, exchange, and coordinated care, with reporting burden and outcomes-focused criticism.

E. Care.Data

  • Care.Data proposed integrating health and social-care information for analysis by NHS and external users, including researchers, managers, planners, academic institutions, and commercial organizations.Its stated aims included understanding disease and treatment, monitoring safety, planning services, and comparing care quality.Source: E. Care.Data · p. 10
  • Care.Data’s opt-out process, unclear limits on data use, and pseudonymized identifiers prompted concerns about consent, access, and re-identification.In October 2014, a government review found major issues with project definition, schedule, budget, quality, and benefits delivery.Source: E. Care.Data · p. 10

F. Incentive Programs

  • HITECH funds adoption of health IT and nationwide EHR capability as a basis for population health management and improved care quality, safety, and efficiency.Its staged program ties incentives to demonstrated implementation and use, with possible penalties for failing to reach stages.Source: F. Incentive Programs · p. 10
  • HITECH-related requirements have drawn criticism for increasing reporting burden while emphasizing reporting requirements rather than outcomes.Later stages include additional quality reporting, clinical decision support, and security risk analysis.Source: F. Incentive Programs · p. 10
  • New York’s DSRIP program offered more than $8 billion to 25 local provider networks, which agreed to share patient data and coordinate care toward a 25% reduction in avoidable hospitalizations.The networks selected targeted programs based on community health assessments.Source: F. Incentive Programs · p. 11
  • The paper argues that population-health risk stratification may reach more people than precision medicine focused on a few patients in clinical trials.The cited example contrasts possible intervention for scattered Type 1 diabetes trial participants with addressing undiagnosed Type 2 diabetes.Source: F. Incentive Programs · p. 11

G. Data Privacy, De-identification and HIPAA

The section links growing patient participation and digital traces to new opportunities for population-level analysis, while emphasizing privacy and reliability concerns. It describes regulatory protections alongside risks from re-identification and secondary use of data.

G. Data Privacy, De-identification and HIPAA

  • 78.8 million people were affected by the 2015 Anthem breach, while weak financial incentives for security help explain the introduction of regulations.HIPAA’s Privacy Rule addresses storage, access, and sharing; its Security Rule sets standards for electronic protected health information, and HITECH adds enforcement penalties.Source: G. Data Privacy, De-identification and… · p. 11

V. THE PATIENT IN CHARGE

  • Patients increasingly manage their health and data, but public traces may expose them to disadvantage, bullying, and social scoring.Source: §V · p. 11

A. Leaving Traces

  • Patients use web resources to research health, compare providers, and connect in condition-specific communities, while clinicians also adopt collaboration tools.Source: A. Leaving Traces · p. 11

B. Analyzing Traces

  • Infodemiology analyzes anonymized online traces for public-health insights, including epidemic detection, but Google Flu Trends’ later fit was poor.Search logs and social media might provide temporal and spatial clues about disease spread, but models have proved difficult.Source: B. Analyzing Traces · pp. 11–12
  • Joint analysis of adverse-event reports, search logs, and social media could improve adverse drug reaction detection, though web-trace quality and bias remain uncertain.Source: B. Analyzing Traces · p. 12
  • Re-identification can expose health status from social or nonmedical traces, enabling risks such as stalking, bullying, and social scoring.Employers may also use third parties to analyze workers’ medical and other personal data; machine learning can infer health conditions from nonmedical information.Source: B. Analyzing Traces · p. 12

C. PatientsLikeMe

The paper describes patient platforms and personal health records as ways to collect, manage, and share health information. It then surveys continuous healthcare technologies, whose potential depends on sensors, connectivity, and data integration.

C. PatientsLikeMe

  • PatientsLikeMe collects quantitative patient data through recurring questionnaires and tools, and sells aggregated, de-identified data to industry partners.Its stated aim is to help partners understand patients’ real-world experiences and disease courses.Source: C. PatientsLikeMe · p. 12

D. Managing Your Own Data

  • A personal health record is a patient-centered assembly of personal health information, but privacy and distribution make PHRs difficult to use in analytics.Patients must be able to store, manage, and control access to their data; rich PHR information may still support personalized advice and alerts.Source: D. Managing Your Own Data · p. 12

VI. CONTINUOUS HEALTHCARE

  • Affordable sensors can shift care from episodic visits toward potentially anytime, anywhere monitoring, with a possible move toward prevention.The paper notes that although many smartphone owners look up health information, only one in five smartphone users own a health app.Source: §VI · p. 13

A. Technological Basis

  • mHealth draws on varied devices and sensor networks to capture health signals, while integration initiatives could help make aggregated data usable for analytics.Smartphone-based lab-on-a-chip technologies may support point-of-care testing, with particular potential in settings with limited infrastructure.Source: A. Technological Basis · p. 13

B. Use Case Types

The use cases span prevention, early detection, disease management, and clinical research, with projects illustrating how mobile technologies collect and apply health data. The evidence described is promising in some areas but remains mixed or incomplete in others.

1) Disease prevention:

  • Mobile tools for cardiovascular prevention include text messages, wearable sensors, and apps, but more evidence is needed despite promising prospects.The American Heart Association encourages further research on mobile technologies for reducing cardiovascular risk behavior.Source: 1) Disease prevention · p. 13

2) Early detection:

  • Early detection can use population-level outbreak signals and individual monitoring, including smart alarms for heart activity, breathing, or falls.A mobile ECG device is described as FDA-cleared and capable of detecting atrial fibrillation.Source: 2) Early detection · p. 13

3) Disease management:

  • Home monitoring may reduce costs by avoiding clinic visits, but trials of mHealth for treatment adherence show mixed evidence and a need for further research.The reviewed applications range from SMS to video messaging and wireless devices.Source: 3) Disease management · pp. 13–14

4) Support of translational research:

  • Mobile studies may let patients choose participation and data sharing, while research apps can continuously measure clinical-trial endpoints alongside physician assessment.Roche’s Parkinson disease app uses six measures based on the Unified Parkinson’s Disease Rating Scale in a Phase I trial.Source: 4) Support of translational research · p. 14

C. Selected Projects

  • Projects combine mobile data and services across settings, including breast cancer research, rural care, services for seniors, and personal self-tracking.The MD Anderson–Polaris pilot combines Apple Watch data with patients’ EHR data; Medic Mobile supports rural care in 23 countries, while Japan Post, IBM, and Apple are designing services for seniors.Source: C. Selected Projects · p. 14

D. Implications for the Clinical Setting and the Doctor’s Office

The paper describes virtual consultations as a possible response to pressure on primary care, while noting that evidence of their effects remains limited. It also surveys regulatory questions and the data-driven development of precision medicine.

D. Implications for the Clinical Setting and the Doctor’s Office

  • Some patients find virtual visits cheaper and more convenient, and demand exists, but evidence is still needed on effects on costs and health outcomes.About 55% of surveyed patients said they would send a skin photo to a dermatologist for consultation.Source: D. Implications for the Clinical… · p. 14

E. Regulatory Implications

  • Mobile medical apps raise questions about FDA jurisdiction and evolving guidance, while reports of unexpected behavior and adverse events underscore the need for stakeholder dialogue.The paper notes about 52 adverse-event reports in the FDA database for one diabetes app over two years.Source: E. Regulatory Implications · pp. 14–15

F. Conclusion

  • The clinical impact of mHealth remains an open question alongside challenges in privacy, sustained user interest, business models, and data integration.Source: F. Conclusion · p. 15

A. Precision Medicine is Changing Healthcare

  • Precision medicine groups patients by disease risk or treatment response to tailor care, building on molecular data and analytics to seek clinically actionable insights.The paper describes growing data volume and variety, the prospect of redefining disease categories through biological causes, and relevance beyond oncology to multiple medical domains.Source: A. Precision Medicine is Changing… · p. 15

B. Understanding Disease on a Molecular Level

The section traces how disease complexity and expanding molecular measurement reshape diagnosis and treatment, from genetic and other omics profiles to biomarker-guided and repurposed therapies.

B. Understanding Disease on a Molecular Level

  • Common diseases often reflect interacting genetic, acquired, and environmental factors, contributing to “missing heritability” and complicating disease explanation.Alzheimer’s is described as involving genetic, lifestyle, and environmental factors; fewer than 5 percent of cases are attributed to specific genetic changes that virtually guarantee disease.Source: B. Understanding Disease on a… · pp. 15–16
  • Germline DNA studies assess inherited variation, while tumor biopsies reveal acquired mutations and amplifications linked to prognosis or therapy response.When biopsy is not standard care for some tumors, circulating tumor cells and circulating tumor DNA provide alternative sources of tumor information.Source: B. Understanding Disease on a… · pp. 15–16
  • RNA sequencing and microarrays capture gene-expression snapshots, while protein and other omics measurements extend molecular characterization beyond DNA.The clinical relevance of epigenomics, metabolomics, and lipidomics remains under investigation.Source: B. Understanding Disease on a… · p. 16
  • Molecular profiles can distinguish cancer subgroups and inform treatment selection; breast cancer markers such as BRCA1, BRCA2, estrogen receptor, progesterone receptor, and HER2 are established examples,,.The section also describes tumor heterogeneity and cancer stem-cell and clonal-evolution models as challenges for personalized treatment.Source: B. Understanding Disease on a… · p. 16

C. Molecular Diagnostics and Drug Therapy

  • Biomarkers measure patient molecules to assess health and can identify targets for therapies tailored to specific molecular variations.Companion diagnostics are developed alongside drugs to help select or exclude patient groups based on expected efficacy and safety.Source: C. Molecular Diagnostics and Drug… · p. 17
  • Drug repurposing applies existing drugs to new indications, and prior safety testing can reduce toxicology-related failure risk compared with traditional drug development,.The section cites a BRAF-targeted drug active in other cancers when the relevant mutation is present.Source: C. Molecular Diagnostics and Drug… · p. 17

D. Implementing Precision Medicine

Precision medicine depends on molecular research at large scale and on integrating its data into clinical care. Its implementation also raises substantial technical, social, ethical, legal, and data-management challenges.

D. Implementing Precision Medicine

  • Implementation spans technology, social and ethical concerns, legal implications, and education, with one insurer beginning to cover sequencing of cancer patients’ full germline and tumor genomes.A single genome can comprise up to 400 GB of data.Source: D. Implementing Precision Medicine · p. 17

E. Big Data in Molecular Research

  • Large molecular consortia combine data from hundreds of thousands of individuals, while studying interacting disease causes and spatio-temporal progression requires still larger samples.The OncoArray study genotyped more than 400,000 people for more than 570,000 variants; breast cancer GWAS identified around 100 risk genes.Source: E. Big Data in Molecular Research · p. 18
  • Population-scale molecular initiatives are expected to yield lessons about data management, emerging standards, and privacy topics such as informed consent.Examples include the Genomics England 100K project and the NIH Precision Medicine Initiative.Source: F. Digitization Challenges in… · p. 18, E. Big Data in Molecular Research · p. 18

F. Digitization Challenges in Precision Medicine

  • A major digitization challenge is connecting omics data to clinical outcomes; meaningful integration requires structured data and synchronized terminology across repositories.CancerLinQ aims to match patient data with outcomes from cases across the United States, drawing on longitudinal records, cohort analyses, quality metrics, reporting, and text analytics.Source: F. Digitization Challenges in… · p. 18

G. Traditional IT Players are Entering Precision Medicine

The section surveys established technology companies’ roles in precision-medicine data infrastructure and describes a future of patient-specific cancer immunotherapy. That vision relies on large-scale sequencing and analysis and faces significant compute and data-management demands.

G. Traditional IT Players are Entering Precision Medicine

  • IT companies are entering precision medicine through partnerships and platforms for clinical decision support, genomic analysis, cloud storage, and data management.Examples include SAP with ASCO’s CancerLinQ, IBM with the New York Genome Center, Dell with TGen, and cloud offerings from Microsoft, AWS, and Google.Source: G. Traditional IT Players are Entering… · pp. 18–19

H. A View to the Future: a Truly “n=1”-Medicine

  • Identifying patient-specific neoepitopes uses tumor-versus-normal whole-genome or exome sequencing, with expression confirmation to reduce false-positive errors.This approach searches beyond the more than 99% of the genome deemed untargetable or unimportant by panel sequencing.Source: H. A View to the Future: a Truly… · p. 19
  • Engineered adenoviruses can deliver neoepitopes to immune cells, and published results describe tumor-associated antigen delivery in late-stage colorectal cancer patients,.The section also describes CAR-engineered NK-92 cells as a potential patient-specific immunotherapy approach.Source: H. A View to the Future: a Truly… · pp. 19–20
  • Neoepitope discovery and targeting require substantial compute, storage, and bandwidth to process sequencing and modeling data in actionable time for each patient.The described infrastructure includes HIPAA-compliant cloud compute and storage and long-term storage for multiple biopsies per patient.Source: H. A View to the Future: a Truly… · p. 20

I. Outlook

The outlook anticipates large-scale data collection, continuous monitoring, and learning healthcare systems, while emphasizing unresolved questions about implementation, privacy, legal responsibilities, and business models. The conclusion frames stakeholder trust and visible benefits as important to realizing more effective care.

I. Outlook

  • Worldwide personalized medicine would require unprecedented-scale data management, major computing and storage investments, and broad stakeholder education and involvement.Privacy and safety concerns also need to be addressed, and the public needs to understand the potential benefits of personalized medicine using Big Data and patient profiling.Source: §I · p. 20

VIII. ASSESSMENTS AND CONCLUSIONS

  • The conclusion presents the EHR as a potential central patient record linking molecular profiles and clinical outcomes to personalized treatment decisions and a learning healthcare system.It says completeness and accuracy are prerequisites for deriving interventional causal conclusions.Source: §VIII · p. 20
  • Continuous healthcare could generate real-time data streams for chronic-patient monitoring, continuous screening, and early intervention.Trusted data centers might also support reminders, alarms, health advice, and preventive measures for individuals and families.Source: §VIII · p. 20
  • The timing and nature of continuous healthcare and personalized medicine’s impact remain unclear, and privacy, liability, legal concerns, and viable data-driven business models are unresolved.The paper also notes that legal frameworks are unclear in many countries and reimbursement models need development.Source: §VIII · pp. 20–21
  • The paper argues that caregivers and patients need to experience benefits, not only added reporting burdens, for the vision of more efficient healthcare to gain stakeholder support.It surveys efforts involving more effective processes, standardized formats, trusted data exchange, and policy and reimbursement changes.Source: §VIII · p. 21
Loading 1606.08075v3…