Sarcomas are a rare and heterogeneous group of malignant tumors arising from mesenchymal tissue, encompassing over 70 distinct histological subtypes with highly variable biology, treatment requirements, and prognoses. Their rarity, combined with their clinical complexity, means that patients typically traverse multiple healthcare settings before receiving a definitive diagnosis. A patient may start at a primary care clinic, undergo initial imaging at a community radiology center, receive a biopsy at a regional hospital, and eventually reach a specialized sarcoma center, each step generating data that remain locked within separate institutional systems.
The data vortex: The authors describe this phenomenon as a "data vortex," a systemic state in which critical clinical information is scattered across siloed electronic health records (EHRs) that cannot communicate with one another. Diagnostic imaging reports, pathology findings, operative notes, radiation therapy plans, and molecular profiling results all coexist in isolation. The result is repeated testing, miscommunications between care teams, and delays in treatment initiation, each of which independently worsens patient outcomes in a disease category where early, accurate intervention is paramount.
Value-Based Healthcare and why sarcoma lags behind: Value-Based Healthcare (VBHC) is a framework that ties reimbursement and care quality to measurable patient outcomes rather than the volume of services delivered. The core principle is that a healthcare system delivers value when it achieves the best possible outcomes per unit of cost. Implementing VBHC for sarcoma is obstructed by the very fragmentation described above: without a complete, longitudinal view of each patient's journey, it is impossible to measure outcomes consistently, benchmark costs accurately, or identify which interventions are truly driving value.
This 2025 paper, published in the Journal of Personalized Medicine by Fuchs, Heesen, and Zhou from the Swiss Sarcoma Network (SSN), introduces ShapeHub, a digital interoperable platform designed to centralize, harmonize, and analyze sarcoma patient data across care settings. The authors use the analogy of a logistics tracking system: just as supply chain software follows a package through each stage of delivery, ShapeHub follows each patient through each point of care, ensuring no data are lost or inaccessible at critical decision moments.
ShapeHub is designed around a structured, directly analyzable internal data model that minimizes the need for external format conversions. The platform's data ingestion architecture follows three principles: "click and collect" for structured clinical inputs entered directly by care providers; "talk and capture" for voice-to-text and natural language processing (NLP)-based extraction from dictated reports and clinical notes; and "scan and sync" for processing PDF documents from external institutions using an AI ontology algorithm that screens for key clinical parameters.
Interoperability standards: For communication with external systems, ShapeHub utilizes Fast Healthcare Interoperability Resources (FHIR), the internationally recognized standard for healthcare data formats that enables structured data exchange between disparate EHR systems. Alongside FHIR, the platform leverages Health Information Exchanges (HIEs), which serve as the actual data exchange conduits between institutions. The authors draw an analogy to GS1 logistics middleware, where standardized identifiers and communication protocols allow goods to be tracked seamlessly across different logistics networks. In healthcare, this translates to a patient's complete record being accessible to any authorized provider within the care network regardless of which system generated it.
Patient identification and privacy: Access management is built around a trackable patient identification system based on World ID, which functions as a secure, cross-institutional patient identifier. This is the healthcare equivalent of a shipment tracking number, linking all patient interactions across institutions under a single, authenticated record. The platform is also planned to incorporate blockchain technology to secure and authenticate each data entry, ensuring information integrity cannot be compromised as records move between systems. For multi-institutional AI model training, ShapeHub employs federated learning, a decentralized approach that trains models across multiple centers without sharing raw patient data, ensuring compliance with GDPR and analogous data privacy regulations.
AI and NLP capabilities: Beyond data aggregation, ShapeHub's AI layer transforms unstructured clinical text into structured, queryable data. NLP algorithms process radiology reports, pathology narratives, and clinical notes to extract key diagnostic and prognostic parameters in real time. Machine learning models trained on multi-omics data, including genomics, proteomics, imaging features, and clinical histories, are being developed within the platform to support personalized treatment recommendations. The platform's AI also enables dynamic risk stratification, comparing individual patient profiles against the accumulated institutional dataset to flag outliers, identify bottlenecks in the care pathway, and propose actionable adjustments.
The authors situate ShapeHub within a landscape of existing digital care coordination and oncology data platforms, identifying both the contributions and limitations of each. This positioning exercise is important for establishing what is genuinely novel about the ShapeHub approach rather than simply restating the idea of EHR integration.
Integrated Practice Units and VBHC models: Porter and Teisberg's VBHC framework gave rise to Integrated Practice Units (IPUs) at institutions like the Cleveland Clinic and MD Anderson Cancer Center, where multidisciplinary teams are co-located to coordinate decision-making and measure outcomes. However, IPUs depend on physical co-location of specialists, making them inherently unscalable for rare diseases whose patient populations are geographically dispersed across regional and national networks. ShapeHub extends the IPU concept by achieving virtual integration through data rather than geography, enabling the same multidisciplinary coordination without requiring patients or specialists to be in the same building.
Oncology-specific EHR platforms: Flatiron Health and ASCO's CancerLinQ are widely used platforms that extract real-world evidence from de-identified oncology EHR datasets. While valuable for population-level research, these platforms operate largely within single-vendor ecosystems and are not optimized for real-time clinical decision support in rare diseases. ShapeHub's differentiation lies in its disease-agnostic modular architecture and its emphasis on interoperability across heterogeneous institutional systems rather than requiring adoption of a single EHR vendor.
National HIEs and rare disease networks: National health data platforms such as France's Dossier Médical Partagé and the Netherlands' MedMij provide longitudinal patient records but primarily serve administrative rather than clinical decision-support functions, with limited integration of AI-driven analytics. The European Reference Networks (ERNs) for rare diseases promote cross-border expert collaboration but lack a shared interoperable data infrastructure for care pathway tracking. ShapeHub can complement ERNs by providing the technical backbone for harmonized data exchange across their member centers. IBM Watson for Oncology and similar AI clinical decision support tools, while pioneering, were siloed to single modalities (typically treatment recommendation) without the end-to-end pathway integration that ShapeHub provides.
The first concrete application of ShapeHub described in the paper involves a retrospective analysis of diagnostic delays within the Swiss Sarcoma Network (SSN). The study analyzed 1,028 patients presented to the multidisciplinary tumor board between 2018 and 2021, making it one of the more substantial real-world datasets for sarcoma diagnostic pathway research.
Measured delays: The analysis decomposed total diagnostic delay into component intervals. The patient interval, representing the time from symptom onset to first medical contact, had a median duration of 2.3 months and was the single largest contributor to overall diagnostic delay. The secondary care interval, representing the time spent between the first specialist referral and a confirmed sarcoma diagnosis at a specialized center, was the most significant component of the diagnostic interval itself. Together, these intervals create a compounding delay that, in a disease where tumor size at diagnosis strongly predicts outcome, translates into measurable harm.
Predictors of longer delays: The analysis identified three independent predictors of prolonged diagnostic intervals: older patient age, axial tumor localization (spine, pelvis, chest wall), and larger tumor size at presentation. Axial sarcomas faced a particularly striking disparity: the median total diagnostic delay for axially localized tumors was 6.7 months compared to 4.1 months for extremity sarcomas, a statistically significant difference. This likely reflects the tendency to attribute axial pain and mass lesions to more common conditions such as degenerative disc disease or benign lipomas.
How ShapeHub addresses this: By implementing real-time referral tracking, ShapeHub enables dynamic identification of which step in the care pathway is generating the bottleneck for a given patient. When a patient's referral stalls at the secondary care interval beyond a defined threshold, the system can trigger alerts to the referring physician and the specialist center simultaneously. This proactive monitoring approach converts a retrospective quality metric into a real-time intervention capability, enabling earlier detection and corrective action within individual patient journeys rather than only identifying systemic problems after the fact through aggregate analysis.
The second case study addresses one of the most consequential and preventable adverse events in sarcoma surgery: the "whoops" surgery, or unplanned excision (UE). A whoops surgery occurs when a soft tissue mass is excised by a surgeon who did not suspect sarcoma before the procedure, without prior staging, specialist consultation, or multidisciplinary planning. Because these surgeries are performed without the margins or staging that a planned sarcoma resection would include, they frequently leave residual tumor and compromise subsequent curative surgery.
Measured outcomes in the SSN cohort: Within the ShapeHub-integrated SSN dataset, 19.6% of all sarcoma patients had undergone an unplanned excision prior to their referral to the specialized center. This figure aligns with rates reported in other national sarcoma registries, suggesting the problem is systemic rather than institution-specific. Using target trial emulation (TTE), a methodological framework that retrospectively compares different treatment strategies across patient cohorts in a manner analogous to a randomized controlled trial, the SSN team compared outcomes between patients who had undergone UE and those whose first resection was a planned procedure at a specialized center.
Risk of local recurrence: Patients who had undergone UE had a 2.27-fold increased risk of local recurrence compared to those undergoing planned resections (95% CI, 1.12-4.60). Residual tumor was found at re-excision or in the re-excised specimen in 31% to 74% of cases following UE, which is the primary driver of this elevated local failure rate. Importantly, despite the higher local recurrence risk, no statistically significant differences were observed in metastasis-free survival, cancer-specific survival, or overall survival between the two groups during the follow-up period, suggesting that prompt re-excision at a specialized center can partially mitigate the long-term oncologic consequences of an initial whoops surgery.
Preventive function of ShapeHub: The platform contributes to reducing UE risk by consolidating available patient data before any surgical decision is made. When a referring surgeon uploads imaging or pathology to the platform, the AI layer can flag features suggestive of sarcoma and recommend specialist consultation before excision. This creates a proactive safety net that intercepts suspicious cases at the community level before an unplanned resection occurs, rather than managing consequences after the fact.
The third case study from the SSN demonstrates how ShapeHub's real-time data integration and treatment outcome tracking enabled the evaluation and adoption of an accelerated preoperative radiotherapy (RT) protocol for soft tissue sarcoma. The standard preoperative RT regimen for extremity and truncal soft tissue sarcomas is normofractionated, typically delivering 50 Gy in 25 fractions of 2 Gy each over five weeks. The SSN piloted an ultrahypofractionated alternative delivering 25 Gy in just 5 fractions of 5 Gy each.
Treatment timeline and outcomes: The median interval between completion of radiation therapy and surgical resection under the ultrahypofractionated protocol was only 16 days, substantially shorter than the 4-8 week waiting period typically recommended after conventional fractionation to allow wound healing optimization. Despite this compressed timeline, the clinical outcomes were compelling: R0 (microscopically clear) resection margins were achieved in 92% of patients, and early treatment-related toxicity was limited to grade 0-1 dermatitis only. No severe radiation-related wound complications were observed in the early follow-up period, which has historically been the primary concern with preoperative RT regimens.
Comparative effectiveness through ShapeHub: ShapeHub enabled a direct, real-world comparison of the ultrahypofractionated protocol against the institutional normofractionated experience. The analysis demonstrated equivalent 2-year local control rates between the two schedules, alongside similar low wound healing complication rates. This comparative effectiveness analysis, conducted prospectively within the platform, provided the level of evidence needed to support adoption of the shorter schedule. The practical implications are significant: a 5-session regimen reduces patient burden by 80% compared to a 25-session schedule, decreases occupancy of radiotherapy units, and reduces operational costs without compromising oncologic outcomes.
Replicability across institutions: The authors emphasize that ShapeHub's ability to prospectively compare outcomes across different hospitals, surgical groups, and care modalities makes such protocol evaluations feasible at a network level without requiring formal randomized controlled trials. The platform's structured data capture ensures that inter-institutional comparisons are conducted on standardized variables, reducing the confounding that typically limits retrospective multi-center analyses.
A central premise of VBHC is that cost and outcome must be measured simultaneously and at the patient level. ShapeHub is being built to support this through integrated cost mapping, where financial data from each stage of the care pathway are captured alongside clinical outcomes. The authors describe a framework for distinguishing cost expenses (direct treatment costs such as imaging, surgery, and radiotherapy) from cost revenues (reimbursements and savings realized through avoided redundancies) and aligning both against patient-level outcome metrics.
Benchmarking findings: Preliminary analyses conducted within the SSN produced a counterintuitive finding: the direct costs of surgical procedures themselves are broadly consistent across hospitals, but overhead costs associated with hospitalization vary significantly between institutions. The implication is that efforts to improve cost efficiency in sarcoma care should focus not on surgical technique standardization but on reducing perioperative hospitalization overhead, which represents the primary cost driver. ShapeHub's integrated financial and clinical data architecture is positioned to identify these discrepancies and create actionable benchmarks for resource allocation decisions.
Digital twin technology: Looking further ahead, the authors describe digital twins as the next frontier in ShapeHub-enabled precision medicine. A digital twin is a virtual computational representation of an individual patient's physiological state and disease trajectory, built from their accumulated clinical, molecular, and imaging data. By simulating the effects of different treatment interventions on the digital twin before they are applied to the real patient, clinicians could theoretically identify the optimal treatment sequence with a much higher degree of confidence than current evidence-based guidelines allow. ShapeHub's harmonized longitudinal dataset is intended to provide the structured, high-quality data infrastructure required to train and validate such models for sarcoma patients.
Scalability beyond sarcoma: The platform architecture is explicitly designed to generalize beyond sarcoma. The modular, disease-agnostic data structure means that the same framework could be applied to other rare and complex cancers, or indeed to any condition characterized by fragmented multi-institutional care and heterogeneous patient populations. The authors position ShapeHub as a reusable blueprint for digital transformation across rare disease networks, with sarcoma serving as the proof-of-concept disease site.
The authors articulate a broad set of implications for ShapeHub's model that extend well beyond individual institutional efficiency. At the health system level, the platform aligns with the concept of a learning health system, a model in which real-world clinical data are continuously fed back into care processes to improve quality and efficiency in a self-reinforcing loop. ShapeHub's ability to incorporate real-time patient feedback, perform target trial emulation across patient groups, and continuously update care pathway analytics supports this adaptive model in a way that static EHR data repositories cannot.
Policy and reimbursement alignment: The integration of cost and outcome data within a single platform creates the infrastructure for outcome-based reimbursement contracts and bundled payments for rare cancers, models that both EU policymakers and the US Centers for Medicare and Medicaid Services (CMS) have been advocating for as alternatives to fee-for-service. ShapeHub's transparent cost-outcome mapping could facilitate the design and implementation of performance-based contracts between payers and specialized sarcoma centers, tying payment to verified patient outcomes rather than to the number of procedures performed.
Patient safety and empowerment: By proactively identifying risk factors for adverse events such as whoops surgeries and undue diagnostic delays, ShapeHub functions as a patient safety system in addition to a data management tool. The authors also note that integrated, up-to-date records give patients themselves greater transparency into their care journey, reducing redundant procedures and supporting informed decision-making through better access to their own clinical history.
Limitations and what remains to be demonstrated: The paper is explicit that ShapeHub remains in an early implementation phase, and the evidence base for its outcomes consists primarily of retrospective analyses from a single national network. The three case studies, while clinically compelling, do not constitute randomized controlled trial-level evidence for the platform's impact. The blockchain and World ID components described for data security and patient identification are planned features rather than fully deployed systems. Scalability to larger, more heterogeneous networks with different regulatory frameworks, EHR vendor landscapes, and languages remains to be demonstrated. The authors call on healthcare institutions, technology providers, and policymakers to collaborate in developing and scaling such systems to build the evidence base needed for broader adoption.