kth.sePublications
7891011121310 of 17
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Perspectives on designing data-driven approaches in healthcare based on real-world evidence
KTH, School of Engineering Sciences in Chemistry, Biotechnology and Health (CBH), Biomedical Engineering and Health Systems, Health Informatics and Logistics.ORCID iD: 0000-0002-3398-2296
2024 (English)Doctoral thesis, comprehensive summary (Other academic)
Sustainable development
SDG 3: Good Health and Well-Being
Abstract [en]

According to the United States Food and Drug Administration, real-world data are defined as “data relating to patient health status and/or the delivery of health care routinely collected from a variety of sources”, and consequently real-world evidence as “the clinical evidence about the usage and potential benefits or risks of a medical product derived from analysis of the real-world data”. In the context of pharmacology and drug development, real-world data and real-world evidence are gaining interest. The main reason is that these can potentially address the limitations of conventional studies according to best practices of the evidence-based medicine paradigm, from which the prioritization of internal control within experimental settings are often done at the expense of generalization to larger populations treated in routine clinical practice. Instead, from a wider multidisciplinary context, it is relevant to pose questions regarding how real-world data can be used to solve the challenges that health-care systems are facing, such as sensitivity to increasing demand of care, and lack of preparedness to sudden changes and lack of resilience.

Despite the acknowledged potential, real-world data poses a series of challenges due to the purpose for which it is collected, primarily for operational purposes rather than for generating evidence, unlike randomized controlled clinical trials. Healthcare systems have been studied as complex adaptive social systems, where the intricate interactions between sub-systems and actors (e.g., patients and doctors) are challenging to capture in data, often introducing confounders that impact the reliability of data. Therefore, a critical discussion on how to leverage data and the data-driven approaches for generating real-world evidence is essential.

The objective of this thesis is to investigate key elements to account for when designing data-driven approaches aimed to overcome practical challenges of real-world data, and generate real-world evidence. The use of data-driven approaches for analyzing real-world data is studied by exploring how real-world processes are reflected in the collected information, pre-processing the data, selecting the appropriate analytical methods, discussing the model logic in relation to the data, interpreting and generating potential clinical insights based on the outcomes, all with feedback from clinical experts.

The thesis investigates these aspects addressing one central research question, subdivided into three specific research questions, through a multi-case study. The three case studies included in this thesis are: small cell lung cancer treatment, emulation of control arms of clinical trials, and flow management of an overcrowded emergency department and hospital wards. The multi-case study approach facilitated a learning process on a case-by-case basis, contributing valuable insights to each individual case. Subsequently, the patterns in the findings from the case studies were aggregated and harmonized into the general framework for designing data-driven approaches for real-world evidence.

The key aspect of this framework is the position of real-world evidence in an intersection between analytical science (the execution of the data-driven approach), with design science (i.e., the discussion with the clinical experts on data information, modeling design, and generation of new clinical insights) to overcome the limits of pure empirical approaches when encountering practical challenges related to real-world data.

The outcomes suggest that the following key aspects should be accounted for when designing data-driven approaches to achieve real-world evidence in health-care: i) the investigation of discrepancies between the real-world processes and the reported information in the data, ii) verification that information and the designed models lead to results that follow a medical logic, iii) involvement of the clinical experts in all the steps of the analysis, iv) healthcare problems should not be solved using empirical approaches alone, but appropriate approaches can generate relevant insights and inform future studies, v) the main purpose of the analysis should be to provide relevant insights for the improvement of clinical practice, and not be limited to the specific case study, but aim to create a learn-confirm cycle, building on clinical knowledge over time and data readiness for secondary purpose.

In conclusion, we must acknowledge the current limitations of the use of real-world data, but at the same time have a constructive vision on learning about what can be achieved now and what could be achieved in the future, by continuously improving design of data-driven approaches.

Abstract [sv]

Enligt United States Food and Drug Administration definieras real-world data som, data relaterat till en patients hälsostatus och/eller tillhandahållandet av hälso- och sjukvård som rutinmässigt samlas in från olika källor. Följaktligen definieras real-world evidens som, klinisk evidens angående använding och potentiella effekter eller risker relaterade till användningen av en medicinsk produkt baserat på analys av real-world data.

Intresset för real-world data och real-world evidens har ökat inom farmakologi och läkemedelsutveckling under de senaste åren. Huvudsakliga anledningen till detta är att dessa adresserar nuvarande begränsningar med konventionell studiedesign med grund i evidens-baserad medicin, där intern kontroll är prioriterad framför förmågan att kunna generalisera resultat till den bredare patientpopulationen som behandlas i klinisk praxis. Från ett multi-disciplinärt perspektiv så är det även relevant att undersöka hur real-world data kan användas för att informera de utmaningar som hälso- och sjukvården nu står inför, såsom ökat patienttryck och låg beredskap för plötsliga förändringar i vårdbehov. Trots potentialen med real-world data finns det ett antal utmaningar med dess användning. Detta beror på att data primärt samlas in för att informera drift av hälso- och sjukvården snarare än att generera evidens till skillnad från randomiserade kontrollerade studier. Hälso- och sjukvården kan beskrivas som ett komplext adaptivt socialt system med komplexa samband mellan undersystem och aktörer (t.ex. patienter och läkare). Dessa komplexa samband är svåra att studera utifrån den data som samlas in vilket snedvrider resultat och tolkningar. Därav krävs vidare forskning kring hur data kan användas tillsammans med data-baserad modellering för att generera real-world evidens från real-world data.

Syftet med denna avhandling är att undersöka vilka de relevanta faktorerna är för design av data-drivna metoder för generering av real-world evidens. Detta utforskas genom att undersöka hur väl data beskriver underliggande processer, bearbetning av real-world data, val av dataanalysmetod, jämförelse av modeller och underliggande processer, tolking och generering av potentiella kliniska fynd, med kliniska experter involverade i alla steg av analysen. Avhandlingen besvarar huvudfrågeställningen och bifrågor med hjälp av tre fallstudier. Dessa är: småcellig lungcancer, emulering av kontrollarmar för randomiserade kontrollerade kliniska studier, och hantering av patientflödet vid en akutavdelning och tillhörande vårdavdelningar.

De studerade fallen tillåter en lärandeprocess som informerar ett mer generellt modell för hur data-drivna metoder kan designas med syfte att generera realworld evidens. Denna modell positionerar real-world evidens mellan analytiska metoder (såsom dataanalys och -modellering) och design (samskapande med kliniska experter kring data, modellering och generering av kliniska insikter) för att överkomma begränsningarna med renodlade empiriska analysmetoder. Den samlade kunskapen av dessa fallstudier ledde till ett antal rekommendationer kring hur datadrivna metoder ska designas för att uppnå real-world evidens, dessa inkluderar: i) undersökning av skillnader mellan verkliga processer och rapporterad data; ii) verifiering av att inhämtad information och designade analysmetoder följer en medicinsk logik; iii) involvering av kliniska experter genom hela analysprocessen; iv) problem i hälso- och sjukvården kan inte lösas genom en renodlad empirisk metod, dock kan välavvägda analyser informera nya insikter om systemet och framtida studier; v) huvudsyftet med analys av real-world data bör vara att generera relevanta insikter för att förbättra klinisk praxis, inte bara i det enskilda fallet men också att skapa en lärandeprocess över fallstudier för att iterativt förbättra klinisk kunskap och datakvalitetet. Avslutningsvis krävs en medvetenhet kring de begränsningar som analys av realworld data innefattar. Samtidigt krävs en konstruktiv vision för hur lärande och evidens kan genereras i nuläget och i framtiden, genom kontinuerlig förbättring av data och data-drivna analysmetoder.

Place, publisher, year, edition, pages
KTH Royal Institute of Technology, 2024. , p. 68
Series
TRITA-CBH-FOU ; 2024:59
Keywords [en]
Real-world data, real-world evidence, oncology, clinical trials design, emergency medicine, data science, healthcare information systems, public health
National Category
Medical and Health Sciences
Research subject
Technology and Health
Identifiers
URN: urn:nbn:se:kth:diva-357636ISBN: 978-91-8106-142-0 (print)OAI: oai:DiVA.org:kth-357636DiVA, id: diva2:1920037
Public defence
2025-01-27, T2, Hälsovägen 11C, via Zoom: https://kth-se.zoom.us/j/63344888971, Huddinge, 13:00 (English)
Opponent
Supervisors
Note

QC 2024-12-10

Available from: 2024-12-10 Created: 2024-12-10 Last updated: 2024-12-16Bibliographically approved
List of papers
1. A novel analytical framework for risk stratification of real‐world data using machine learning: A small cell lung cancer study
Open this publication in new window or tab >>A novel analytical framework for risk stratification of real‐world data using machine learning: A small cell lung cancer study
Show others...
2022 (English)In: Clinical and Translational Science, ISSN 1752-8054, E-ISSN 1752-8062, Vol. 15, no 10, p. 2437-2447Article in journal (Refereed) Published
Abstract [en]

In recent studies, small cell lung cancer (SCLC) treatment guidelines based on Veterans’ Administration Lung Study Group limited/extensive disease staging and resulted in broad and inseparable prognostic subgroups. Evidence suggests that the eight versions of tumor, node, and metastasis (TNM) staging can play an important role to address this issue. The aim of the present study was to improve the detection of prognostic subgroups from a real-word data (RWD) cohort of patients and analyze their patterns using a development pipeline with thoracic oncologists and machine learning methods. The method detected subgroups of patients informing unsupervised learning (partition around medoids) including the impact of covariates on prognosis (Cox regression and random survival forest). An analysis was carried out using patients with SCLC (n = 636) with stage IIIA–IVB according to TNM classification. The analysis yielded k = 7 compacted and well-separated clusters of patients. Performance status (Eastern Cooperative Oncology Group-Performance Status), lactate dehydrogenase, spreading of metastasis, cancer stage, and CRP were the baselines that characterized the subgroups. The selected clustering method outperformed standard clustering techniques, which were not capable of detecting meaningful subgroups. From the analysis of cluster treatment decisions, we showed the potential of future RWD applications to understand disease, develop individualized therapies, and improve healthcare decision making.

Place, publisher, year, edition, pages
Wiley, 2022
Keywords
C reactive protein; carboplatin; cisplatin; etoposide; irinotecan; lactate dehydrogenase; platinum complex
National Category
Medical and Health Sciences Cancer and Oncology
Identifiers
urn:nbn:se:kth:diva-321440 (URN)10.1111/cts.13371 (DOI)000832654100001 ()35856401 (PubMedID)2-s2.0-85135121261 (Scopus ID)
Funder
Swedish Cancer Society, CAN2021/1469 Pj01Swedish Cancer Society, CAN 2018/597KTH Royal Institute of Technology
Note

QC 20221115

Available from: 2022-11-14 Created: 2022-11-14 Last updated: 2024-12-10Bibliographically approved
2. Exploring the discrepancies between clinical trials and real-world data: A small-cell lung cancer study
Open this publication in new window or tab >>Exploring the discrepancies between clinical trials and real-world data: A small-cell lung cancer study
Show others...
2024 (English)In: Clinical and Translational Science, ISSN 1752-8054, E-ISSN 1752-8062, Vol. 17, no 8, article id e13909Article in journal (Refereed) Published
Abstract [en]

The potential of real-world data to inform clinical trial design and supplement control arms has gained much interest in recent years. The most common approach relies on reproducing control arm outcomes by matching real-world patient cohorts to clinical trial baseline populations. However, recent studies pointed out that there is a lack of replicability, generalisability, and consensus. In this article, we propose a novel approach that aims to explore and examine these discrepancies by concomitantly investigating the impact of selection criteria and operations on the measurements of outcomes from the patient data. We tested the approach on a dataset consisting of small-cell lung cancer patients receiving platinum-based chemotherapy regimens from a real-world data cohort (n = 223) and six clinical trial control arms (n = 1224). The results showed that the discrepancy between real-world and clinical trial data potentially depends on differences in both patient populations and operational conditions (e.g., frequency of assessments, and censoring), for which further investigation is required. Discovering and accounting for confounders, including hidden effects of differences in operations related to the treatment process and clinical trial study protocol, would potentially allow for improved translation between clinical trials and real-world data. Continued development of the method presented here to systematically explore and account for these differences could pave the way for transferring learning across clinical studies and developing mutual translation between the real-world and clinical trials to inform clinical study design.

Place, publisher, year, edition, pages
Wiley, 2024
National Category
Cancer and Oncology
Identifiers
urn:nbn:se:kth:diva-351885 (URN)10.1111/cts.13909 (DOI)001285427300001 ()39113428 (PubMedID)2-s2.0-85200661148 (Scopus ID)
Note

QC 20240829

Available from: 2024-08-19 Created: 2024-08-19 Last updated: 2024-12-10Bibliographically approved
3. Real-World Evidence on Dose-Reduction and Treatment Outcomes in Small Cell Lung Cancer; A Bayesian mixed effects and Competitive Risk Approach
Open this publication in new window or tab >>Real-World Evidence on Dose-Reduction and Treatment Outcomes in Small Cell Lung Cancer; A Bayesian mixed effects and Competitive Risk Approach
Show others...
(English)Manuscript (preprint) (Other academic)
Abstract [en]

Background: Small cell lung cancer (SCLC) is a challenging disease to treat due to rapid progression, development of chemo- resistance, and discrepancies in outcomes between real-world data and clinical trials. Previous studies lack comprehensive analyses of intermediate events and the treatment process, such as treatment decisions, progression of disease, and the occurrence of adverse events (AEs) over time. The aim of this study was to apply advanced statistical methods on a longitudinal SCLC data set in order to identify factors of importance for the risk of AEs and for survival.

Methods: Information was collected on the treatment pathways of 421 SCLC patients treated at the Karolinska University Hospital between 2016-2022 (Stockholm, Sweden). Analysis focused on the impact of dose-adjustment on adverse events (AEs), including neutropenia, by estimating odds ratios (OR) using Bayesian mixed-effects modelling. Covariate’s effects on ECOG performance status (PS) deterioration and early discontinuation of chemotherapy with cause-specific hazard ratios (csHR) were explored using competitive risk models. This approach was applied to cohorts of patients receiving first line platinum etoposide, and second line platinum etoposide or platinum irinotecan.

Results: At the end of the first line treatment, most patients exhibited tumour regression (n=167). Patients with neutropenia had longer overall survival (HR: 0.70 [0.53, 0.92]). Higher etoposide dose levels were associated with subsequent occurrences of AEs (OR: 5.97 [1.41, 30.5]) and neutropenia (OR: 3.55 [1.03, 13.3]). Dose adjustment did not affect overall survival as long as the patient completed the four-dose regimen treatment. For the second-line, fewer patients completed four treatment cycles and the most common reason of early discontinuation was tumour progression (n=72). Male patients experienced fewer AEs and better first line treatment response compared to females (csHR: 0.51 [0.25, 0.90]). High-risk patients (here defined as ECOG PS 2-3, or age over 75 years) with early discontinuation of therapy had survival outcomes similar to those who did not receive therapy.

Conclusions: Our results indicate that SCLC therapies may benefit from more individualized dosing strategies. These strategies would aim to balance improved survival with reduced risk of AEs, particularly neutropenia. It would also be beneficial to assess the risk-benefit of treating specific subgroups, including patients receiving second line therapy. Real-world data are crucial for studying therapy response and risk-benefit of treating patient groups that are underrepresented in clinical trials.

National Category
Cancer and Oncology
Research subject
Medical Technology
Identifiers
urn:nbn:se:kth:diva-357277 (URN)10.1101/2024.12.04.24318490 (DOI)
Note

QC 20241206

Available from: 2024-12-06 Created: 2024-12-06 Last updated: 2024-12-10Bibliographically approved
4. Diagnosing an overcrowded emergency department from its Electronic Health Records
Open this publication in new window or tab >>Diagnosing an overcrowded emergency department from its Electronic Health Records
Show others...
2024 (English)In: Scientific Reports, E-ISSN 2045-2322, Vol. 14, no 1, article id 9955Article in journal (Refereed) Published
Abstract [en]

Emergency department overcrowding is a complex problem that persists globally. Data of visits constitute an opportunity to understand its dynamics. However, the gap between the collected information and the real-life clinical processes, and the lack of a whole-system perspective, still constitute a relevant limitation. An analytical pipeline was developed to analyse one-year of production data following the patients that came from the ED (n = 49,938) at Uppsala University Hospital (Uppsala, Sweden) by involving clinical experts in all the steps of the analysis. The key internal issues to the ED were the high volume of generic or non-specific diagnoses from non-urgent visits, and the delayed decision regarding hospital admission caused by several imaging assessments and lack of hospital beds. Furthermore, the external pressure of high frequent re-visits of geriatric, psychiatric, and patients with unspecified diagnoses dramatically contributed to the overcrowding. Our work demonstrates that through analysis of production data of the ED patient flow and participation of clinical experts in the pipeline, it was possible to identify systemic issues and directions for solutions. A critical factor was to take a whole systems perspective, as it opened the scope to the boundary effects of inflow and outflow in the whole healthcare system.

Place, publisher, year, edition, pages
Nature Research, 2024
National Category
Medical and Health Sciences
Identifiers
urn:nbn:se:kth:diva-346372 (URN)10.1038/s41598-024-60888-9 (DOI)001225890200015 ()38688997 (PubMedID)2-s2.0-85191754138 (Scopus ID)
Note

QC 20240617

Available from: 2024-05-14 Created: 2024-05-14 Last updated: 2024-12-10Bibliographically approved

Open Access in DiVA

fulltext(20006 kB)103 downloads
File information
File name FULLTEXT01.pdfFile size 20006 kBChecksum SHA-512
18800940076963077dcfe45d2caacc269bd555b73557e4b18d2ce294192771252bbeb1eeb4a9aae506dfd81a91206cc762b1b398aa79b8c23cba8885e735e3d5
Type fulltextMimetype application/pdf

Authority records

Marzano, Luca

Search in DiVA

By author/editor
Marzano, Luca
By organisation
Health Informatics and Logistics
Medical and Health Sciences

Search outside of DiVA

GoogleGoogle Scholar
Total: 103 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

isbn
urn-nbn

Altmetric score

isbn
urn-nbn
Total: 867 hits
7891011121310 of 17
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf