Digital health technologies in oncology services: A scoping review of clinical implementation, organisational evidence gaps, and policy implications

May combine two claims.

Digital literacy, language, access to devices, reliable internet connectivity, and confidence using technology affected participation across several studies.

“Digital literacy, language, access to devices, reliable internet connectivity, and confidence using technology affected participation across several studies [25,37,46,48,50,58].”

• cites [25] Sinha Gregory, 2023, The feasibility, acceptability, and usability of telehealth visits, Front. Med. (Lausanne). sourcePartly supported

  The Results (provider survey; page not provided) identify lack of needed technology and poor internet connectivity as challenges to video visits. The Discussion also links low technical skills and comfort with technology to telehealth difficulties. Our copy does not identify language as a participation barrier, so it supports only part of the multi-citation sentence. (LLM, conf=0.84)

  Experimental — accuracy not yet measured
• cites [37] Sánchez, 2024, Addressing the unmet needs of women with breast cancer in Mexico: a non-randomised pilot study of the digital ePRO intervention, BMJ Open. sourcePartly supported

  The Results (Research area II, individual-level barriers; page unavailable) report that low digital literacy made the ePRO app difficult to use and led some women to seek help. The Methods (Quantitative stage; page unavailable) require internet access through a phone, computer or tablet for eligibility, and the Results (Research area I; page unavailable) describe the value of mobile-data support to some participants. This study does not report language barriers or show that device access, reliable connectivity, and confidence using technology each affected participation. (LLM, conf=0.87)

  Experimental — accuracy not yet measured
• cites [46] Schuler, 2016, Implementation of a mobile inpatient quality of life (QoL) assessment for oncology nursing, Support. Care Cancer sourceNot checked — no full text of the source
• cites [48] Yılmaz, 2022, Enhancing patient participation of older migrant cancer patients: needs, barriers, and eHealth, Ethn. Health sourcePartly supported

  The Language barrier and low health literacy section identifies language as a barrier to participation. The Technology acceptance section reports difficulty using the Health Communicator and doubts about being able to use it, supporting the digital-skills and confidence portions. The cited paper does not establish that access to devices or reliable internet connectivity affected participation. (LLM, conf=0.87)

  Experimental — accuracy not yet measured
• cites [50] Neo, 2024, Enhancing palliative care for advanced cancer patients: evaluating implementation and impact of a virtual nurse-led symptom monitoring and telehealth initiative, BMC Palliat. Care sourcePartly supported

  The Discussion (Main findings; page not available) says technological literacy was correlated with the telemedicine service’s reach. Results (Reach; page not available) also lists IT-related difficulties among reasons for nonparticipation. The paper does not establish that language, device access, reliable internet connectivity, or confidence using technology affected participation. (LLM, conf=0.86)

  Experimental — accuracy not yet measured
• cites [58] Smith, 2022, Exploring cancer patients’, caregivers’, and clinicians’ utilisation and experiences of telehealth services during COVID-19: A qualitative study, Patient Educ. Couns. sourcePartly supported

  Results, Theme 6, identifies low technological literacy, lack of confidence, and poor connectivity as barriers to video telehealth. Results, Theme 3, describes difficulties using interpreters for non-English-speaking patients. The paper discusses hospital-provided equipment, but the passages found do not establish that patients’ access to devices affected participation. (LLM, conf=0.84)

  Experimental — accuracy not yet measured

May combine two claims.

Previous systematic reviews in oncology have generally examined a single category of technology or concentrated on specific user groups, most often patients and nurses.

“Previous systematic reviews in oncology have generally examined a single category of technology, such as artificial intelligence (AI) applications in cancer nursing [11], telephonebased symptom management [12], patient empowerment platforms [13], large language models (LLMs) for clinical decision-making [14], or telehealth during the COVID-19 pandemic [15], or have concentrated on specific user g”

• cites [11] O’Connor, 2024, The application and use of artificial intelligence in cancer nursing: A systematic review, Eur. J. Oncol. Nurs. sourceSupported
• cites [12] Ream, 2020, Telephone interventions for symptom management in adults with cancer, Cochrane Database Syst. Rev. 2020 sourceSupported
• cites [16] Sezgin, 2025, The Effectiveness of Patient-Centered Digital Empowerment Programs in Hematological Cancer Care: A Systematic Review and Meta-Analysis of Randomized Controlled Trials, World Evid. Based Nurs. sourceSupported

May combine two claims.

Digital tools could add to clinician workload when alerts were excessive, outputs were not actionable, or responsibilities for responding were unclear.

“Several studies showed that digital tools could add to clinician workload when alerts were excessive, outputs were not actionable, or responsibilities for responding were unclear [23,45,50,55].”

• cites [23] Basch, 2020, Clinical Utility and User Perceptions of a Digital System for Electronic Patient-Reported Symptom Monitoring During Routine Cancer Care: Findings From the PRO-TECT Trial, JCO Clin. Cancer Inform. sourcePartly supported

  The Discussion (page unavailable) says frequent alerts added to nurses’ existing workload, and 47% of nurses felt there were too many. The paper does not link added workload to unactionable outputs; it generally describes the information as useful. Its Digital ePRO System and Questions section says clinical responses were not prescribed, but does not establish that responsibility for responding was unclear. (LLM, conf=0.86)

  Experimental — accuracy not yet measured
• cites [45] Schmalz, 2020, Digital Monitoring and Management of Patients With Advanced or Metastatic Non-Small Cell Lung Cancer Treated With Cancer Immunotherapy and Its Impact on Quality of Clinical Care: Interview and Survey Study Among Health Care Professionals and Patients, J. Med. Internet Res. sourcePartly supported

  The Results section (‘Exploration of HCP Needs, Expectations, and Perceived Value’) reports that workload was the most prominent challenge and that managing the standalone tool and entering data sometimes offset time savings. It links the extra time chiefly to interoperability and system-integration issues, not excessive alerts, nonactionable outputs, or unclear responsibility for responding. Page numbers are unavailable in our copy. (LLM, conf=0.88)

  Experimental — accuracy not yet measured
• cites [50] Neo, 2024, Enhancing palliative care for advanced cancer patients: evaluating implementation and impact of a virtual nurse-led symptom monitoring and telehealth initiative, BMC Palliat. Care sourceCould not verify

  The Methods (Intervention description; page unavailable) describe a nurse reviewing symptom surveys and triaging patients, but do not assess whether the tool increased clinician workload. The Discussion (Main findings; page unavailable) addresses patient survey completion and access barriers, not excessive alerts, unactionable outputs, or unclear response responsibilities. Searches for those terms and clinician workload found no supporting passage in our copy. (LLM, conf=0.76)

  Experimental — accuracy not yet measured
• cites [55] Girgis, 2020, Web-Based Patient-Reported Outcome Measures for Personalized Treatment and Care (PROMPT-Care): Multicenter Pragmatic Nonrandomized Trial, J. Med. Internet Res. sourcePartly supported

  The Discussion (Strengths and Limitations; page unavailable) says alerts were limited to reduce false positives and that balancing screening burden against timely alerts is challenging. But the Discussion (Integration and Clinical Use; page unavailable) says this trial’s low alert volume fit existing workflows without onerous additional work. The paper does not report nonactionable outputs or unclear response responsibilities; its Methods instead describe recommended actions and a designated reviewer. (LLM, conf=0.78)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Two included studies originated from Asia.

“From Asia, two studies were identified”

May combine two claims.

These technologies were used mainly for symptom monitoring, communication, follow-up, survivorship support, clinical decision support, and care coordination.

“These technologies were used mainly for symptom monitoring, communication, follow-up, survivorship support, clinical decision support, and care coordination.”

May combine two claims.

Reported benefits included earlier symptom detection, improved communication, greater patient reassurance, and more structured clinical workflows.

“Reported benefits included earlier symptom detection, improved communication, greater patient reassurance, and more structured clinical workflows.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Several studies were conducted in Europe (n = 12).

“Several studies were conducted in Europe (n = 12)”

Probably not a substantial claim: may not pass S3 (load-bearing).

The 41 included studies were published between 2016 and 2025.

“The 41 included studies were published between 2016 and 2025”

Probably not a substantial claim: may not pass S3 (load-bearing).

The largest proportion of included studies (n = 19) originated from North America.

“the largest proportion (n = 19) originated from North America”

May combine two claims.

Evidence involving non-clinical professionals and AI-enabled implementation remains limited.

“Because evidence involving non-clinical professionals and AI-enabled implementation remains limited”

Probably not a substantial claim: may not pass S3 (load-bearing).

One included study originated from Turkey.

“one study originated in Turkey”

May combine two claims.

The included studies comprised 11 RCTs, 5 non-randomised quantitative designs, 10 mixed-methods studies, 14 qualitative research, and 1 quantitative descriptive study.

“the studies included randomised controlled trials (RCTs, n = 11), non-randomised quantitative designs (n = 5), mixed-methods studies (n = 10), qualitative research (n = 14), and quantitative descriptive studies (n = 1).”

May combine two claims. The quote only partly states this claim.

ePRO systems were the most common tools, used in 15 of the 41 studies.

“In total, 15 of the 41 studies utilised such systems.”

• Also stated in: § 4.1 Digital Monitoring and ePRO Systems · plain

  “Digital monitoring and electronic patient-reported outcome (ePRO) systems were the most frequently used tools in the included studies.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Digital tools were classified into eleven categories.

“Digital tools were classified into eleven categories.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Seven included studies originated from Australia.

“seven from Australia”

May combine two claims.

Common barriers included limited digital literacy, declining engagement over time, alert burden, workflow misalignment, poor interoperability, and inadequate technical infrastructure.

“Common barriers included limited digital literacy, declining engagement over time, alert burden, workflow misalignment, poor interoperability, and inadequate technical infrastructure.”

May combine two claims.

Patients reported better understanding of anatomy, immobilisation, and radiation delivery, as well as greater confidence and less anxiety than with conventional education methods.

“Patients reported better understanding of anatomy, immobilisation, and radiation delivery, as well as greater confidence and less anxiety than with conventional education methods [56].”

May combine two claims.

Tools such as COMPASS integrated these data with clinician dashboards, offering opportunities for more personalised follow-up and earlier intervention.

“Tools such as COMPASS integrated these data with clinician dashboards, offering opportunities for more personalised follow-up and earlier intervention [28].”

May combine two claims.

Telehealth raised challenges, including digital literacy barriers, poor internet access, falling engagement over time, difficulty conducting physical examinations remotely, and limitations in sensitive or emotionally complex consultations.

“However, telehealth also raised challenges, including digital literacy barriers, poor internet access, falling engagement over time, difficulty conducting physical examinations remotely, and limitations in sensitive or emotionally complex consultations [24,25,31,50,58].”

May combine two claims.

ePRO systems improved communication with HCPs, enabled earlier intervention, enhanced reassurance during treatment, and supported self-management through tailored feedback and educational resources.

“For patients, ePRO systems improved communication with HCPs, enabled earlier intervention, enhanced reassurance during treatment, and supported self-management through tailored feedback and educational resources [23,29,37,43,45,59].”

May combine two claims.

Mobile applications supported patient education, symptom self-management, treatment adherence, survivorship support, sexual health, and clinical decision support.

“Their functions ranged from patient education and symptom self-management to treatment adherence, survivorship support, sexual health, and clinical decision support [21,39,41,51,53].”

May combine two claims.

OIN provided accessible evidence-based information for patients and nurses, reducing anxiety while also supporting nursing education and professional networking.

“OIN provided accessible evidence-based information for patients and nurses, reducing anxiety while also supporting nursing education and professional networking [36].”

May combine two claims.

Concerns persisted around bias, privacy, accountability, cost, workforce displacement, and the limited ability of AI systems to provide empathy or emotional support.

“Concerns also persisted around bias, privacy, accountability, cost, workforce displacement, and the limited ability of AI systems to provide empathy or emotional support [49].”

May combine two claims.

For HCPs, mobile applications improved decision-making, efficiency, and monitoring, although their value depended heavily on integration into routine practice.

“For HCPs, mobile applications improved decision-making, efficiency, and monitoring, although their value depended heavily on integration into routine practice.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

For NCPs, telehealth and telemonitoring were viewed as tools that could improve flexibility, productivity, care coordination, and longer-term planning.

“For NCPs, telehealth and telemonitoring were viewed as tools that could improve flexibility, productivity, care coordination, and longer-term planning [8,38].”

May combine two claims.

Education and navigation platforms aimed to improve patient knowledge, support shared decision-making, and strengthen continuity of care.

“Education and navigation platforms aimed to improve patient knowledge, support shared decision-making, and strengthen continuity of care.”

May combine two claims.

Telehealth was associated with improved QoL, greater reassurance, reduced travel, and fewer ED visits and hospital admissions.

“Additional reported benefits included improved QoL, greater reassurance, reduced travel, and fewer ED visits and hospital admissions [25,30,50,52,58].”

May combine two claims.

Nurse-led telephone support reduced severe toxicities and smaller declines in QoL compared with usual care.

“Nurse-led telephone support reduced severe toxicities and smaller declines in QoL compared with usual care [35].”

May combine two claims.

ePRO systems were associated with more structured consultations, earlier symptom detection, improved efficiency, and more standardised care for HCPs.

“For HCPs, ePRO systems were associated with more structured consultations, earlier symptom detection, improved efficiency, and more standardised care [23,45,54].”

May combine two claims.

The main barriers to ePRO systems were alert burden, poor interoperability with EHRs, inconsistent uptake among oncologists, and limits on nurses' prescribing authority.

“The main barriers were alert burden, poor interoperability with EHRs, inconsistent uptake among oncologists, and limits on nurses' prescribing authority [23,32,37,45,54,55].”

May combine two claims.

HCPs reported that telehealth improved communication with patients and families, supported structured follow-up, and increased satisfaction with care delivery.

“HCPs reported that telehealth improved communication with patients and families, supported structured follow-up, and increased satisfaction with care delivery [10,25,27,45,58].”

May combine two claims.

AI-enabled systems were mainly used to improve timeliness and accuracy in imaging-related tasks and to support interpretation of patient-reported outcomes and triage processes.

“AI-enabled systems were mainly used to improve timeliness and accuracy in imaging-related tasks and to support interpretation of patient-reported outcomes and triage processes [40].”

May combine two claims.

Major weaknesses in current cancer pathways remained evident, including diagnostic delays across several tumour types and high false-positive rates.

“At the same time, major weaknesses in current cancer pathways remained evident, including diagnostic delays across several tumour types and high false-positive rates [40].”

May combine two claims.

These models were associated with higher one-year survival, promising cost-effectiveness, and smoother transitions between hospital and community services, especially for patients living alone.

“These models were associated with higher one-year survival, promising cost-effectiveness, and smoother transitions between hospital and community services, especially for patients living alone [42].”

May combine two claims.

Recurring barriers to mobile apps included digital literacy problems, navigation difficulties, engagement fatigue, technical problems, limited educational scope, poor interoperability, additional workload, data overload, cultural resistance to mobile-device use in clinical settings, and uncertainty about long-term scalability.

“However, recurring barriers included digital literacy problems, navigation difficulties, engagement fatigue, technical problems, limited educational scope, poor interoperability, additional workload, data overload, cultural resistance to mobile-device use in clinical settings, and uncertainty about long-term scalability [28,29,39,41,42,51,53].”

May combine two claims.

Many included studies were early-phase pilot or feasibility studies, so evidence on long-term effectiveness, sustainability, and scalability remains limited.

“Many were early-phase pilot or feasibility studies, so evidence on long-term effectiveness, sustainability, and scalability remains limited.”

May combine two claims.

The included studies were highly heterogeneous in design, populations, technologies, and outcomes, limiting direct comparison.

“The included studies were highly heterogeneous in design, populations, technologies, and outcomes, limiting direct comparison.”

May combine two claims.

Only three included studies explicitly involved non-clinical professionals, preventing firm conclusions about managerial use, procurement, resource allocation, or system-level implementation.

“Only three included studies explicitly involved non-clinical professionals, preventing firm conclusions about managerial use, procurement, resource allocation, or system-level implementation.”

• AssumptionRests on an assumption: Involving non-clinical professionals is necessary to generate evidence about organisational adoption, resource allocation, procurement, or system-level implementation.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

ePRO, telehealth, and mobile apps are the most common digital oncology tools.

“• ePRO, telehealth, and mobile apps are the most common digital oncology tools”

May combine two claims.

Barriers include EHR integration challenges, IT downtime, and variable staff access to systems.

“Barriers: EHR integration challenges, IT downtime, variable staff access to systems.”

May combine two claims.

Digital literacy gaps, workload, and poor interoperability are recurring barriers.

“• Digital literacy gaps, workload, and poor interoperability are recurring barriers”

May combine two claims.

Evidence concerning AI governance, workforce preparation, scalability, sustainability, and economic impact was less developed and should inform future implementation research rather than definitive policy requirements.

“In contrast, evidence concerning AI governance, workforce preparation, scalability, sustainability, and economic J o u r n a l P r e -p r o o f impact was less developed and should inform future implementation research rather than definitive policy requirements.”

May combine two claims.

Across clinical and non-clinical settings, digital health technologies were used mainly to support treatment delivery, follow-up, and survivorship care, with more limited but emerging applications in diagnostic processes and administrative management.

“Third, across clinical and non-clinical settings, digital health technologies were used mainly to support treatment delivery, follow-up, and survivorship care, with more limited but emerging applications in diagnostic processes and administrative management.”

May combine two claims.

The evidence base was small, rapidly evolving, and frequently focused on technical or early-stage evaluation rather than routine clinical implementation.

“The evidence base was small, rapidly evolving, and frequently focused on technical or early-stage evaluation rather than routine clinical implementation.”

Classified as probably not a substantial claim. May combine two claims.

AI governance and workforce preparation are relevant emerging concerns, but the recommendations that can be drawn from the included studies remain limited.

“AI governance and workforce preparation are relevant emerging concerns, but the recommendations that can be drawn from the included studies remain limited.”

Classified as probably not a substantial claim.

Two reviewers independently screened titles and abstracts, as well as full texts, against eligibility criteria.

“two reviewers (TT, CC) independently screened titles and abstracts, as well as full texts, against eligibility criteria.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The protocol was initially registered in PROSPERO as a systematic review (CRD420251074190).

“The protocol was initially registered in PROSPERO as a systematic review (CRD420251074190).”

May combine two claims.

The evidence was concentrated on patient-facing and clinical applications, with substantially less attention to organisational and non-clinical workforce perspectives.

“However, the evidence was concentrated on patient-facing and clinical applications, with substantially less attention to organisational and non-clinical workforce perspectives.”

• Also stated in: § 4. Discussion · boosted

  “the evidence was strongly concentrated on patient-facing applications and frontline clinical practice.”

Classified as probably not a substantial claim.

A narrative synthesis was used instead of meta-analysis because of heterogeneity in interventions and outcomes.

“Because of heterogeneity in interventions and outcomes, a narrative synthesis was used instead of meta-analysis”

Classified as probably not a substantial claim. May combine two claims.

The search was conducted in PubMed, Embase, CINAHL, and Web of Science (2015-June 2025), complemented by hand-searching reference lists.

“A comprehensive search was conducted in PubMed, Embase, CINAHL, and Web of Science (2015-June 2025), complemented by hand-searching the reference lists.”

May combine two claims.

Mobile applications can support both symptom management and broader survivorship needs.

“Together, these studies suggest that mobile applications can support both symptom management and broader survivorship needs.”

May combine two claims.

Poor interoperability reduced the clinical usefulness of otherwise promising technologies and contributed to duplicate documentation, fragmented information, and additional workload.

“Poor interoperability reduced the clinical usefulness of otherwise promising technologies and contributed to duplicate documentation, fragmented information, and additional workload.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

The adoption of digital technologies depends not only on clinical effectiveness but also on organisational decisions concerning procurement, workflow redesign, staffing, infrastructure, interoperability, financing, and long-term sustainability.

“This distinction is important because the adoption of digital technologies depends not only on clinical effectiveness but also on organisational decisions concerning procurement, workflow redesign, staffing, infrastructure, interoperability, financing, and long-term sustainability.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Conflicts in screening were resolved by consensus or, if required, by a third reviewer.

“Conflicts were resolved by consensus or, if required, by a third reviewer (KM).”

Probably not a substantial claim: may not pass S3 (load-bearing).

The eligibility criteria were developed using the PICOS framework and adapted for the scoping purpose of this review.

“The eligibility criteria were developed using the PICOS framework and adapted for the scoping purpose of this review [19].”

• cites [19] Amir-Behghadami, 2020, Population, Intervention, Comparison, Outcomes and Study (PICOS) design as a framework to formulate eligibility criteria in systematic reviews, Emerg. Med. J. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S2 (checkable).

AI governance needs more study.

“• Interoperability is the strongest policy priority; AI governance needs more study”

Classified as probably not a substantial claim.

Interoperability is the strongest policy priority.

“• Interoperability is the strongest policy priority; AI governance needs more study”

May combine two claims.

The review was restricted to English-language, peer-reviewed publications and excluded grey literature, which may have introduced publication and language bias.

“The review was restricted to English-language, peer-reviewed publications and excluded grey literature, which may have introduced publication and language bias.”

Classified as probably not a substantial claim.

By including qualitative, quantitative, mixed-methods, pilot, and implementation studies, the review provides a broad picture of how digital health technologies are described and used in oncology care.

“By including qualitative, quantitative, mixed-methods, pilot, and implementation studies, it provides a broad picture of how digital health technologies are described and used in oncology care.”

Classified as probably not a substantial claim. May combine two claims.

Older adults, rural populations, and people facing language or literacy barriers may require alternative access routes, technical assistance, or non-digital options.

“Older adults, rural populations, and people facing language or literacy barriers may require alternative access routes, technical assistance, or non-digital options.”

May combine two claims.

The review cannot provide a comprehensive account of organisational or managerial use, and findings concerning service planning, resource allocation, and system-level implementation should be interpreted cautiously.

“The review therefore cannot provide a comprehensive account of organisational or managerial use, and findings concerning service planning, resource allocation, and systemlevel implementation should be interpreted cautiously.”

• AssumptionRests on an assumption: Involving non-clinical professionals is necessary to generate evidence about organisational adoption, resource allocation, procurement, or system-level implementation.

Classified as probably not a substantial claim. May combine two claims.

No formal risk-of-bias assessment was conducted, so the findings should be interpreted as a descriptive mapping rather than a graded evaluation of evidence strength.

“As is typical in scoping reviews, no formal risk-of-bias assessment was conducted, so the findings should be interpreted as a descriptive mapping rather than a graded evaluation of evidence strength.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The review followed transparent methods, including PROSPERO registration and adherence to PRISMA-ScR guidance, and used a comprehensive search across multiple databases.

“It followed transparent methods, including PROSPERO registration and adherence to PRISMA-ScR guidance, and used a comprehensive search across multiple databases.”