Pith. sign in

REVIEW 4 major objections 3 minor 76 references

Voice-based AI Agents: Filling the Economic Gaps in Digital Health Delivery

T0 review · 4 major / 3 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Voice-based AI agents, driven by large language models, can economically fill the gap in continuous patient monitoring where human care is too costly, and a 33-patient pilot indicates most patients accept and some prefer the AI.

desk verdict A readable industry/academia viewpoint with new pilot preference data, but the headline cost-savings conclusion is an assumed inequality, not a measured result. read the letter →

arxiv 2507.16229 v1 pith:TS6ZERIR submitted 2025-07-22 cs.AI cs.CYcs.ETcs.HCcs.SE

Bo Wen , Chen Wang , Qiwei Han , Raquel Norel , Julia Liu , Thaddeus Stappenbeck , Jeffrey L. Rogers This is my paper
classification cs.AIcs.CYcs.ETcs.HCcs.SE
keywords voice-basedAIagentslargelanguagemodelsremotepatientmonitoringdigitalhealthhealthcareeconomicscost-utilityanalysisengagementequity
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that voice-based AI agents powered by large language models can economically fill the gap in continuous patient monitoring between clinical visits. The authors propose a cost-utility model that reserves physicians, nurses, and caregivers for higher-severity cases, and assigns routine monitoring of low-severity patients (severity below a threshold $S_l$) to AI voice systems, which they argue is the only economically viable way to deliver continuous preventive care at scale. As evidence, they report a pilot in which 33 patients with inflammatory bowel disease used a telephone-based AI assistant (Agent PULSE) for health assessments; 70% expressed acceptance of AI-driven monitoring, and 37% preferred it over the group-based alternative. The economic case matters because chronic disease monitoring is currently labor-bound, and voice is the one interface available to nearly every patient regardless of device ownership, literacy, or broadband access.

What carries the argument

The load-bearing objects are the two cost equations and the severity-threshold allocation model. Equation (1), $E = (C_h - C_a)/C_h \times 100\%$, defines the percentage cost saving of AI monitoring; Equation (2), $ICER = (C_a - C_h)/(QALY_a - QALY_h)$, incorporates quality-adjusted life years so that savings can be judged against health outcomes. The allocation model stratifies patients by disease severity $S$ with thresholds $S_h$, $S_m$, $S_l$: specialized physician care above $S_h$, nursing care between $S_m$ and $S_h$, untrained caregivers between $S_l$ and $S_m$, and AI monitoring for $S \leq S_l$—the 'blue zone' where human-delivered care is economically unjustifiable but monitoring still helps. The empirical carrier is Agent PULSE, a telephone-based AI assistant that runs natural-language health assessments, converts free-form speech into structured questionnaire responses through an automated analysis framework, and escalates concerning cases to human providers.

What would settle it

A randomized trial assigning chronic disease patients to either LLM voice check-ins or routine nurse telephone follow-ups, tracking hospital readmissions, undetected deterioration events, survey completion, and quality-adjusted life years over 6–12 months, would settle the claim: if the AI arm shows worse detection or adherence, the $E > 0$ savings are not realizable while maintaining service levels.

Watch

Extended reading notes

Core claim

The paper's central claim is that LLM-powered voice assistants are the economically justified 'entry point' for preventive care and continuous monitoring in the low-severity range, formalized by the cost-efficiency ratio $E = (C_h - C_a)/C_h \times 100\%$, where $C_h$ is the cost of human-provided care and $C_a$ the cost of AI-powered intervention; when $E > 0$, AI yields savings 'while maintaining service levels.' The pilot of Agent PULSE, a telephonic LLM assistant, is presented as empirical support: 70% of the 33 inflammatory bowel disease patients accepted the AI modality, 37% preferred it over the group-based sessions, and the response-completeness data show patients disclose most about daily activities and symptoms while holding back on more sensitive items. The authors conclude that AI voice agents can extend care reach, reduce per-patient monitoring costs, and potentially reduce hospital readmissions by keeping patients stable during mild periods, while freeing human staff for higher-acuity work.

Load-bearing premise

The savings claim in Equation (1) assumes an AI check-in maintains the same service level as a human check-in, but the pilot measures patient preference only—not health outcomes, adherence, or whether the AI detects deterioration as reliably as a nurse would.

Editorial extensions

If this is right

  • Chronic disease monitoring programs could move routine check-ins from nurses to AI voice systems, freeing clinical staff for higher-acuity cases and reducing per-patient monitoring costs.
  • Because it works over ordinary telephone lines, voice-AI monitoring could reach patients without smartphones or reliable internet, directly addressing access barriers for older, low-income, and rural populations.
  • The observed acceptance rates suggest patient willingness is not the main barrier to adoption; the binding constraints are response latency, health-record integration, and privacy compliance.
  • If the economic model is used as a planning tool, value-based providers and insurers would have a direct financial incentive to deploy voice-AI monitoring to reduce hospitalizations and emergency visits.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The authors do not extend the argument, but the same cost logic should apply to any chronic condition with a structured symptom questionnaire—diabetes, heart failure, and depression follow-ups are natural testbeds for the framework.
  • The pilot's low completion rates on sensitive questions (as low as 6%) suggest AI may systematically change what patients disclose; if the bias runs toward under-reporting risk, the cost equations would need an outcome penalty that Equation (1) currently lacks.
  • The paper's technical roadmap implies that conversational latency, not clinical accuracy, may be the practical gatekeeper for scale; a direct test would be to measure whether 2–3 times faster responses reduce survey abandonment.
  • The economic logic generalizes beyond voice: wherever a task has near-zero marginal cost once built, the same severity-threshold argument could justify algorithmic triage over human labor in other health-delivery settings.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 3 minor

Summary. The paper argues that LLM-powered voice-based AI agents can fill economic and accessibility gaps in chronic disease monitoring, particularly for underserved populations. It presents a cost-utility model (Eqs. 1-6) intended to show large cost savings for AI monitoring, and it reports a pilot study of Agent PULSE with 33 inflammatory bowel disease patients from Morehouse School of Medicine, in which 70% expressed acceptance of the AI modality and 37% preferred it over alternatives. The paper also describes the system architecture, technical challenges, and policy considerations, and it makes recommendations to healthcare executives, professionals, patients, technologists, and policymakers.

Significance. If the central cost-effectiveness claim were established, this would be a valuable contribution to digital health delivery in resource-constrained settings. The paper has genuine strengths: the Agent PULSE architecture is described in enough detail to be reproduced; the appendix includes a full sample conversation with automated MHBI and EQ-5D-3L extraction; and the authors transparently report the large item-level non-completion rates in Figure 4 and acknowledge the trade-off between authenticity and completeness in Section IV-B. The pilot's feasibility data are useful preliminary results. However, the economic and clinical equivalence claims are not supported by the evidence presented, and the cost-savings conclusion is structurally built into the model's assumptions rather than derived from measured inputs or outcomes.

major comments (4)
  1. [Section III-A, Eq. (1)] The claim that a positive E 'results in cost savings while maintaining service levels' is an assumption, not a result. Equation (1) contains only costs; no quality, safety, adherence, or outcome term enters the definition. The pilot measures stated preference and acceptance only, so the equivalence of service levels between AI and human monitoring is never established. This is load-bearing for every subsequent economic conclusion.
  2. [Section IV-B, Figure 4] The paper's own data completeness analysis shows item-level completion rates ranging from 100% down to 6%, and the text explicitly states that patients interacted differently with the AI and that this 'also resulted in less consistent completion of the full assessment.' Missing responses on sensitive symptom questions directly threaten the premise that AI monitoring preserves the information needed to detect deterioration and trigger escalation, which is exactly the assumption required for the 'while maintaining service levels' claim.
  3. [Section III-B, Eqs. (3)-(5)] The savings conclusion R is guaranteed by construction once Eq. (4) assumes Va ≪ Cm. No pilot cost records, fixed-cost amortization, or sensitivity analysis are provided, so the abstract's claim of 'huge potential savings' is an arithmetic consequence of an unverified inequality rather than an empirical result. The model needs at least a range estimate for Cm, Va, and F, or a breakeven analysis, before any numerical claim can be made.
  4. [Section IV-C] The pilot is a single-site, uncontrolled study of 33 patients with no comparator arm, no pre-registered outcomes, and no statistical tests; it collects no QALY data that would feed Eq. (2). The statement that the pilot 'clearly demonstrated that voice-based AI agents can effectively fill gaps in care delivery' and 'validate key aspects of our economic model' is therefore overstated. At most, the pilot provides preliminary feasibility and acceptability evidence.
minor comments (3)
  1. [Section IV-B, Figure 3] The '70% acceptance' figure includes 18% who valued both approaches and 15% with no strong preference; 'acceptance' should be defined more precisely, since 15% expressing no preference is not equivalent to actively accepting the AI modality.
  2. [Section III-B, Eq. (4)] The notation is inconsistent: the text calls Ca the per-patient cost, but Eq. (4) defines CAI = F + Np × Va, which is a total cost. The units and definitions of Ca, CAI, and Va should be clarified.
  3. [Section IV-B, data completeness analysis] The authors state that transcript examination identified factors such as survey fatigue and environmental distractions, but no qualitative analysis method, coding procedure, or inter-rater reliability is described; a brief methodological note would help readers assess this finding.

Circularity Check

1 steps flagged · score 8.0 of 10

The claimed 'huge potential savings' is an algebraic restatement of the assumed cost inequality Va≪Cm in Eqs. (4)-(5); no measured costs, QALYs, or service-level data enter the model.

  1. self definitional [Section III-B, Eqs. (4)-(5); the same claim appears in the Abstract as 'our cost-utility analysis demonstrates huge potential savings.']
    "For AI-driven voice agents, the per-patient cost Ca is significantly lower due to automation and scalability, leading to: CAI = F + (Np × Va), where Va ≪ Cm. This reflects the economic principle that digital technologies typically have high fixed costs but very low marginal costs [41], creating a distinctly different economic profile compared to human-delivered services."

    Substituting CAI = F + Np*Va into R gives R = 1 - Va/Cm - F/(Np*Cm). Because the model itself declares Va ≪ Cm and treats F as amortizable, R > 0 is not an empirical estimate or a measured prediction; it is the paper's input inequality restated in ratio form. No pilot cost ledger, amortized fixed cost, or external benchmark feeds Eq. (5), and the service-level condition promised in Eq. (1) is absent from the formula. The abstract's 'cost-utility analysis demonstrates huge potential savings' therefore reports a definitional consequence of the model's assumptions, not a derived result.

full rationale

The central economic claim reduces by construction to its own input. Equation (1) defines cost reduction purely as a cost ratio, and the 'while maintaining service levels' condition is asserted in prose rather than represented by any quality or outcome variable. Equations (4)-(5) then make the savings conclusion immediate: CAI is defined with Va ≪ Cm, so the reported 'cost reduction factor' is algebraically forced to be positive before any data are consulted. The 33-patient pilot measures stated acceptance and preference (70% acceptance, 37% preference), not cost, QALYs, adherence, or safety endpoints, and Section IV-B concedes patients 'interacted with the AI system differently' and that this 'also resulted in less consistent completion of the full assessment,' leaving the service-equivalence premise untested. The paper's own Figure 4 shows item-level completion rates as low as 6%, which weighs against assuming equivalent data quality. No self-citation is load-bearing here: the SOLOMON citations concern system components, not the economic equations, so this is not a self-citation-chain problem. The finding is not that the model is false; it is that the model's headline result is a restatement of the assumed inequality Va≪Cm, which is why this is circular rather than merely under-evidenced.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claims rest on unmeasured cost variables and unverified clinical equivalence. No new theoretical entities are introduced; Agent PULSE and SOLOMON are named systems, not postulated constructs.

free parameters (3)
  • AI variable cost per patient (Va)
    Section III.B Eq. (4) assumes Va is far smaller than human monitoring cost Cm; no cost data or unit price is reported anywhere in the paper.
  • Human monitoring cost per patient per month (Cm)
    Used in Eqs. (3)-(5) to compute 'cost savings'; never measured or estimated from actual staffing data.
  • QALY values (QALYa, QALYh)
    Eq. (2) defines ICER using quality-adjusted life years for AI and human care; no QALY estimates are provided, and the narrative treats them as equal without evidence.
assumptions (4)
  • ad hoc to paper Thresholds Sh, Sm, Sl define when physician, nursing, lay caregiver, or AI care is economically appropriate.
    Figure 1 and Section III.A assume these thresholds without calibration to data; the entire resource allocation argument rests on them.
  • ad hoc to paper AI monitoring preserves service quality and health outcomes equal to human care.
    Eq. (1) claims cost savings 'while maintaining service levels'; the pilot does not measure outcomes, only preferences.
  • domain assumption Telephone penetration is universal enough that voice agents reduce access disparities.
    Section II.C argues universal access; the pilot does not test patients who lack phones, internet, or literacy.
  • domain assumption LLM voice agents have high fixed cost F and low variable cost Va.
    Section III.B Eq. (4) models software economics; reasonable but unverified for this deployment.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Voice-based AI Agents: Filling the Economic Gaps in Digital Health Delivery." pith.science (2026). https://pith.science/paper/TS6ZERIR

@misc{pith2026250716229,
  author       = {Pith},
  title        = {Pith review of: Voice-based AI Agents: Filling the Economic Gaps in Digital Health Delivery},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TS6ZERIR}},
  note         = {Machine review of arXiv:2507.16229}
}
read the original abstract

The integration of voice-based AI agents in healthcare presents a transformative opportunity to bridge economic and accessibility gaps in digital health delivery. This paper explores the role of large language model (LLM)-powered voice assistants in enhancing preventive care and continuous patient monitoring, particularly in underserved populations. Drawing insights from the development and pilot study of Agent PULSE (Patient Understanding and Liaison Support Engine) -- a collaborative initiative between IBM Research, Cleveland Clinic Foundation, and Morehouse School of Medicine -- we present an economic model demonstrating how AI agents can provide cost-effective healthcare services where human intervention is economically unfeasible. Our pilot study with 33 inflammatory bowel disease patients revealed that 70\% expressed acceptance of AI-driven monitoring, with 37\% preferring it over traditional modalities. Technical challenges, including real-time conversational AI processing, integration with healthcare systems, and privacy compliance, are analyzed alongside policy considerations surrounding regulation, bias mitigation, and patient autonomy. Our findings suggest that AI-driven voice agents not only enhance healthcare scalability and efficiency but also improve patient engagement and accessibility. For healthcare executives, our cost-utility analysis demonstrates huge potential savings for routine monitoring tasks, while technologists can leverage our framework to prioritize improvements yielding the highest patient impact. By addressing current limitations and aligning AI development with ethical and regulatory frameworks, voice-based AI agents can serve as a critical entry point for equitable, sustainable digital healthcare solutions.

Figures

Figures reproduced from arXiv: 2507.16229 by the authors.

Figure 1
Figure 1. Economic model showing the relationship between disease sever [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Agent PULSE Architecture: The system integrates a voice interface [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 4
Figure 4. Data completeness across different question categories. Questions [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

76 extracted references · 70 canonical work pages

  1. [1]

    Envisioning a better us health care system for all: coverage and cost of care,

    R. Crowley, H. Daniel, T. G. Cooney, L. S. Engel, and Health and Public Policy Committee of the American College of Physicians, “Envisioning a better us health care system for all: coverage and cost of care,” Ann. Intern. Med., vol. 172, no. 2 Supplement, pp. S7–S32, 2020

    work page 2020

  2. [2]

    Public health challenges and responses to the growing ageing populations,

    H. T. Khan, K. M. Addo, and H. Findlay, “Public health challenges and responses to the growing ageing populations,” Public Health Challenges, vol. 3, no. 3, p. e213, 2024

    work page 2024

  3. [3]

    Health care 2020: reengineering health care delivery to combat chronic disease,

    R. V . Milani and C. J. Lavie, “Health care 2020: reengineering health care delivery to combat chronic disease,” Am. J. Med. , vol. 128, no. 4, pp. 337–343, 2015

    work page 2020

  4. [4]

    Covid-19 and health care’s digital revolution,

    S. Keesara, A. Jonas, and K. Schulman, “Covid-19 and health care’s digital revolution,” N. Engl. J. Med. , vol. 382, no. 23, p. e82, 2020

    work page 2020

  5. [5]

    Understanding and responding to health literacy as a social determinant of health,

    D. Nutbeam and J. E. Lloyd, “Understanding and responding to health literacy as a social determinant of health,” Annu. Rev. Public Health , vol. 42, no. 2021, pp. 159–173, 2021

    work page 2021

  6. [6]

    Social determinants, health literacy, and disparities: intersections and controversies,

    D. Schillinger, “Social determinants, health literacy, and disparities: intersections and controversies,” HLRP: Health Literacy Research and Practice, vol. 5, no. 3, pp. e234–e243, 2021

    work page 2021

  7. [7]

    Testing and evaluation of health care applications of large language models: a systematic review,

    S. Bedi, Y . Liu, L. Orr-Ewing, D. Dash, S. Koyejo, A. Callahan, J. A. Fries, M. Wornow, A. Swaminathan, L. S. Lehmann et al. , “Testing and evaluation of health care applications of large language models: a systematic review,” JAMA, 2024

    work page 2024

  8. [8]

    The future landscape of large language models in medicine,

    J. Clusmann, F. R. Kolbinger, H. S. Muti, Z. I. Carrero, J.-N. Eckardt, N. G. Laleh, C. M. L. L ¨offler, S.-C. Schwarzkopf, M. Unger, G. P. Veldhuizen et al. , “The future landscape of large language models in medicine,” Commun. Med., vol. 3, no. 1, p. 141, 2023

    work page 2023

Show all 76 references
  1. [9]

    A multimodal generative ai copilot for human pathology,

    M. Y . Lu, B. Chen, D. F. Williamson, R. J. Chen, M. Zhao, A. K. Chow, K. Ikemura, A. Kim, D. Pouli, A. Patel et al. , “A multimodal generative ai copilot for human pathology,” Nature, vol. 634, no. 8033, pp. 466–473, 2024

    2024

  2. [10]

    Large language models in medicine,

    A. J. Thirunavukarasu, D. S. J. Ting, K. Elangovan, L. Gutierrez, T. F. Tan, and D. S. W. Ting, “Large language models in medicine,” Nat. Med., vol. 29, no. 8, pp. 1930–1940, 2023

    1930

  3. [11]

    Reclaiming voice with ai,

    F. N. Mirza, A. Bogan, A. L. Beam, A. K. Manrai, and R. Ali, “Reclaiming voice with ai,” p. AIp2401000, 2024

    2024

  4. [12]

    Talk2care: An llm-based voice assistant for communication between healthcare providers and older adults,

    Z. Yang, X. Xu, B. Yao, E. Rogers, S. Zhang, S. Intille, N. Shara, G. G. Gao, and D. Wang, “Talk2care: An llm-based voice assistant for communication between healthcare providers and older adults,” Proc. ACM Interact. Mob. Wearable Ubiquitous Technol. , vol. 8, no. 2, pp. 1–35, 2024

    2024

  5. [13]

    Healthcare voice ai assistants: factors influencing trust and intention to use,

    X. Zhan, N. Abdi, W. Seymour, and J. Such, “Healthcare voice ai assistants: factors influencing trust and intention to use,” Proc. ACM Hum.-Comput. Interact., vol. 8, no. CSCW1, pp. 1–37, 2024

    2024

  6. [14]

    V oice-based conversational agents for the prevention and management of chronic and mental health conditions: systematic literature review,

    C. B ´erub´e, T. Schachner, R. Keller, E. Fleisch, F. v Wangenheim, F. Barata, and T. Kowatsch, “V oice-based conversational agents for the prevention and management of chronic and mental health conditions: systematic literature review,” J. Med. Internet Res. , vol. 23, no. 3,...

    2021

  7. [15]

    Use of intelligent voice assis- tants by older adults with low technology use,

    A. Pradhan, A. Lazar, and L. Findlater, “Use of intelligent voice assis- tants by older adults with low technology use,” ACM Trans. Comput.- Hum. Interact., vol. 27, no. 4, pp. 1–27, 2020

    2020

  8. [16]

    Barriers and facilitators to the use of e-health by older adults: a scoping review,

    J. Wilson, M. Heinsch, D. Betts, D. Booth, and F. Kay-Lambkin, “Barriers and facilitators to the use of e-health by older adults: a scoping review,” BMC Public Health , vol. 21, pp. 1–12, 2021

    2021

  9. [17]

    Ubiquitous accessibility for people with visual impairments: Are we there yet?

    S. M. Billah, V . Ashok, D. E. Porter, and I. Ramakrishnan, “Ubiquitous accessibility for people with visual impairments: Are we there yet?” in Proc. 2017 CHI Conference on Human Factors in Computing Systems , 2017, pp. 5862–5868

    2017

  10. [18]

    The impact of voice assistant home devices on people with disabilities: a longitudinal study,

    A. D. Vieira, H. Leite, and A. V . L. V olochtchuk, “The impact of voice assistant home devices on people with disabilities: a longitudinal study,” Technol. Forecast. Soc. Change, vol. 184, p. 121961, 2022

    2022

  11. [19]

    A review of user interface design for interactive machine learning,

    J. J. Dudley and P. O. Kristensson, “A review of user interface design for interactive machine learning,” ACM Trans. Interact. Intell. Syst. , vol. 8, no. 2, pp. 1–37, 2018

    2018

  12. [20]

    V oice-controlled intelligent personal assistants in health care: International delphi study,

    A. Ermolina and V . Tiberius, “V oice-controlled intelligent personal assistants in health care: International delphi study,” J. Med. Internet Res., vol. 23, no. 4, p. e25312, 2021

    2021

  13. [21]

    Exploring how older adults use a smart speaker– based voice assistant in their first interactions: Qualitative study,

    S. Kim et al. , “Exploring how older adults use a smart speaker– based voice assistant in their first interactions: Qualitative study,” JMIR mHealth uHealth, vol. 9, no. 1, p. e20427, 2021

    2021

  14. [22]

    Browsing with alexa: Interrogating the impact of voice assistants as web interfaces,

    S. Natale and H. Cooke, “Browsing with alexa: Interrogating the impact of voice assistants as web interfaces,” Media Cult. Soc. , vol. 43, no. 6, pp. 1000–1016, 2021

    2021

  15. [23]

    A survey on recent advances in llm-based multi-turn dialogue systems,

    Z. Yi, J. Ouyang, Y . Liu, T. Liao, Z. Xu, and Y . Shen, “A survey on recent advances in llm-based multi-turn dialogue systems,” arXiv preprint arXiv:2402.18013, 2024

    2024 arXiv

  16. [24]

    The Death of IVR: AI’s Evolving Impact on CX,

    M. Shah, “The Death of IVR: AI’s Evolving Impact on CX,” Forbes Tech Council, 2024, accessed: March 2025. [Online]. Available: https://www.forbes.com/councils/forbestechcouncil/2024/11/13/the- death-of-ivr-ais-evolving-impact-on-cx/

    2024

  17. [25]

    Understanding the benefits and challenges of deploying conversational ai leveraging large language models for public health intervention,

    E. Jo, D. A. Epstein, H. Jung, and Y .-H. Kim, “Understanding the benefits and challenges of deploying conversational ai leveraging large language models for public health intervention,” in Proc. 2023 CHI Conference on Human Factors in Computing Systems , 2023, pp. 1–16

    2023

  18. [26]

    Role of health information technology in addressing health disparities: patient, clinician, and system perspectives,

    X. Zhang, B. Hailu, D. C. Tabor, R. Gold, M. H. Sayre, I. Sim, B. Jean- Francois, C. A. Casnoff, T. Cullen, V . A. Thomas Jr et al. , “Role of health information technology in addressing health disparities: patient, clinician, and system perspectives,” Med. Care, vol. 57, pp. ...

    2019

  19. [27]

    Understanding barriers and design opportunities to improve healthcare and qol for older adults through voice assistants,

    C. Chen, J. G. Johnson, K. Charles, A. Lee, E. T. Lifset, M. Hogarth, A. A. Moore, E. Farcas, and N. Weibel, “Understanding barriers and design opportunities to improve healthcare and qol for older adults through voice assistants,” in Proc. 23rd International ACM SIGACCESS Con...

    2021

  20. [28]

    A question of access: exploring the perceived benefits and barriers of intelligent voice assistants for improving access to consumer health resources among low-income older adults,

    P. Nallam, S. Bhandari, J. Sanders, and A. Martin-Hammond, “A question of access: exploring the perceived benefits and barriers of intelligent voice assistants for improving access to consumer health resources among low-income older adults,” Gerontol. Geriatr. Med. , vol. 6, p...

    2020

  21. [29]

    A scoping review of patient-facing, behavioral health interventions with voice assistant technology targeting self-management and healthy lifestyle behaviors,

    E. Sezgin, L. K. Militello, Y . Huang, and S. Lin, “A scoping review of patient-facing, behavioral health interventions with voice assistant technology targeting self-management and healthy lifestyle behaviors,” Transl. Behav. Med., vol. 10, no. 3, pp. 606–628, 2020

    2020

  22. [30]

    On the concept of health capital and the demand for health,

    M. Grossman, “On the concept of health capital and the demand for health,” J. Polit. Econ., vol. 80, no. 2, pp. 223–255, 1972

    1972

  23. [31]

    Cost-effectiveness of aducanumab to prevent alzheimer’s disease progression at current list price,

    P. Sinha and J. A. Barocas, “Cost-effectiveness of aducanumab to prevent alzheimer’s disease progression at current list price,”Alzheimer’s Dement., vol. 8, no. 1, p. e12256, mar 2022

    2022

  24. [32]

    Economists in health care: saviors, or elephants in a porcelain shop?

    U. E. Reinhardt, “Economists in health care: saviors, or elephants in a porcelain shop?” Am. Econ. Rev., vol. 79, no. 2, pp. 337–342, 1989

    1989

  25. [33]

    What is value in health care?

    M. E. Porter, “What is value in health care?” N. Engl. J. Med., vol. 363, no. 26, pp. 2477–2481, 2010

    2010

  26. [34]

    Risk adjustment in medicare aco program deters coding increases but may lead acos to drop high-risk beneficiaries,

    A. A. Markovitz, J. M. Hollingsworth, J. Z. Ayanian, E. C. Norton, N. M. Moloci, P. L. Yan, and A. M. Ryan, “Risk adjustment in medicare aco program deters coding increases but may lead acos to drop high-risk beneficiaries,” Health Aff. , vol. 38, no. 2, pp. 253–261, 2019, pMI...

    2019

  27. [35]

    Economies of scope,

    J. C. Panzar and R. D. Willig, “Economies of scope,” Am. Econ. Rev. , vol. 71, no. 2, pp. 268–272, 1981

    1981

  28. [36]

    Strategy of prevention: lessons from cardiovascular disease,

    G. Rose, “Strategy of prevention: lessons from cardiovascular disease,” Br. Med. J., vol. 282, no. 6279, pp. 1847–1851, 1981

    1981

  29. [37]

    Implementing and scaling artificial intelligence: A review, framework, and research agenda,

    N. Haefner, V . Parida, O. Gassmann, and J. Wincent, “Implementing and scaling artificial intelligence: A review, framework, and research agenda,” Technol. Forecast. Soc. Change, vol. 197, p. 122878, 2023

    2023

  30. [38]

    Besanko, D

    D. Besanko, D. Dranove, M. Shanley, and S. Schaefer, Economics of strategy, 7th Edition. John Wiley & Sons, 2017

    2017

  31. [39]

    Production-line approach to service,

    T. Levitt, “Production-line approach to service,” Harv. Bus. Rev., 1972

    1972

  32. [40]

    Uncertainty and the welfare economics of medical care. 1963,

    K. J. Arrow, “Uncertainty and the welfare economics of medical care. 1963,” Bull. World Health Organ. , vol. 82, no. 2, pp. 141–149, 2004

    1963

  33. [41]

    Shapiro and H

    C. Shapiro and H. R. Varian, Information rules: a strategic guide to the network economy. Harvard Business Press, 1998

    1998

  34. [42]

    M. F. Drummond, M. J. Sculpher, K. Claxton, G. L. Stoddart, and G. W. Torrance, Methods for the economic evaluation of health care programmes. Oxford University Press, 2015

    2015

  35. [43]

    The triple aim: care, health, and cost,

    D. M. Berwick, T. W. Nolan, and J. Whittington, “The triple aim: care, health, and cost,” Health Aff., vol. 27, no. 3, pp. 759–769, 2008

    2008

  36. [44]

    Chanoff, M

    M. Chanoff, M. Furst, D. Sabbah, and M. Wegman, The Heart of Innovation: A Field Guide for Navigating to Authentic Demand. Berrett- Koehler Publishers, 2023

    2023

  37. [45]

    Comparative analysis of open- source language models in summarizing medical text data,

    Y . Chen, Z. Wang, and F. Zulkernine, “Comparative analysis of open- source language models in summarizing medical text data,” in Proc. 2024 IEEE International Conference on Digital Health (ICDH) . IEEE, 2024, pp. 126–128

    2024

  38. [46]

    Enhancing reasoning to adapt large language models for domain-specific applications,

    B. Wen and X. Zhang, “Enhancing reasoning to adapt large language models for domain-specific applications,” 2024, adaptive Foundation Models (AFM) Workshop at Neural Information Processing Systems (NeurIPS 2024). [Online]. Available: https://arxiv.org/abs/2502.04384

    2024 arXiv

  39. [47]

    Detection of acute 3, 4- methylenedioxymethamphetamine (mdma) effects across protocols using automated natural language processing,

    C. Agurto, G. A. Cecchi, R. Norel, R. Ostrand, M. Kirkpatrick, M. J. Baggott, M. C. Wardle, H. d. Wit, and G. Bedi, “Detection of acute 3, 4- methylenedioxymethamphetamine (mdma) effects across protocols using automated natural language processing,” Neuropsychopharmacology, vo...

    2020

  40. [48]

    Short dietary assessment instruments,

    National Cancer Institute, “Short dietary assessment instruments,” Website, 2024, accessed: March 2025. [Online]. Available: https://epi.grants.cancer.gov/diet/screeners/

    2024

  41. [49]

    Application of machine learning optimization in cloud computing resource scheduling and management,

    Y . Zhang, B. Liu, Y . Gong, J. Huang, J. Xu, and W. Wan, “Application of machine learning optimization in cloud computing resource scheduling and management,” in Proc. 5th International Conference on Computer Information and Big Data Applications , 2024, pp. 171–175

    2024

  42. [50]

    Existential challenges for healthcare data protection in the united states,

    N. Terry, “Existential challenges for healthcare data protection in the united states,” Ethics Med. Public Health, vol. 3, no. 1, pp. 19–27, 2017

    2017

  43. [51]

    Privacy-aware cloud auditing for gdpr compliance verification in online healthcare,

    M. Barati, G. S. Aujla, J. T. Llanos, K. A. Duodu, O. F. Rana, M. Carr, and R. Ranjan, “Privacy-aware cloud auditing for gdpr compliance verification in online healthcare,” IEEE Trans. Ind. Informat. , vol. 18, no. 7, pp. 4808–4819, 2021

    2021

  44. [52]

    Generative ai in medical practice: in- depth exploration of privacy and security challenges,

    Y . Chen and P. Esmaeilzadeh, “Generative ai in medical practice: in- depth exploration of privacy and security challenges,” J. Med. Internet Res., vol. 26, p. e53008, 2024

    2024

  45. [53]

    Defend architecture: a privacy by design platform for gdpr compliance,

    L. Piras, M. G. Al-Obeidallah, A. Praitano, A. Tsohou, H. Mouratidis, B. Gallego-Nicasio Crespo, J. B. Bernard, M. Fiorani, E. Magkos, A. C. Sanz et al., “Defend architecture: a privacy by design platform for gdpr compliance,” in Proc. Trust, Privacy and Security in Digital Bu...

    2019

  46. [54]

    Framework for healthcare security practice analysis, modeling and incentivization,

    P. K. Yeng, B. Yang, and E. A. Snekkenes, “Framework for healthcare security practice analysis, modeling and incentivization,” in Proc. 2019 IEEE International Conference on Big Data (Big Data) . IEEE, 2019, pp. 3242–3251

    2019

  47. [55]

    Designing personality-adaptive conversational agents for mental health care,

    R. Ahmad, D. Siemon, U. Gnewuch, and S. Robra-Bissantz, “Designing personality-adaptive conversational agents for mental health care,” Inf. Syst. Front., vol. 24, no. 3, pp. 923–943, 2022

    2022

  48. [56]

    Conversational health agents: A personalized llm-powered agent framework,

    M. Abbasian, I. Azimi, A. M. Rahmani, and R. Jain, “Conversational health agents: A personalized llm-powered agent framework,” arXiv preprint arXiv:2310.02374, 2023

    2023 arXiv

  49. [57]

    Rag in health care: a novel framework for improving communication and decision-making by addressing llm limitations,

    K. K. Y . Ng, I. Matsuba, and P. C. Zhang, “Rag in health care: a novel framework for improving communication and decision-making by addressing llm limitations,” NEJM AI , vol. 2, no. 1, p. AIra2400380, 2025

    2025

  50. [58]

    Conversational ai with large language models to increase the uptake of clinical guidance,

    G. Macia, A. Liddell, and V . Doyle, “Conversational ai with large language models to increase the uptake of clinical guidance,” Clin. eHealth, vol. 7, pp. 147–152, 2024

    2024

  51. [59]

    Cachegen: Kv cache compression and streaming for fast large language model serving,

    Y . Liu, H. Li, Y . Cheng, S. Ray, Y . Huang, Q. Zhang, K. Du, J. Yao, S. Lu, G. Ananthanarayanan et al., “Cachegen: Kv cache compression and streaming for fast large language model serving,” in Proc. ACM SIGCOMM 2024 Conference , 2024, pp. 38–56

    2024

  52. [60]

    Keep the cost down: A review on methods to optimize llm’s kv-cache consumption,

    L. Shi, H. Zhang, Y . Yao, Z. Li, and H. Zhao, “Keep the cost down: A review on methods to optimize llm’s kv-cache consumption,” arXiv preprint arXiv:2407.18003, 2024

    2024 arXiv

  53. [61]

    Cacheblend: Fast large language model serving for rag with cached knowledge fusion,

    J. Yao, H. Li, Y . Liu, S. Ray, Y . Cheng, Q. Zhang, K. Du, S. Lu, and J. Jiang, “Cacheblend: Fast large language model serving for rag with cached knowledge fusion,” arXiv preprint arXiv:2405.16444 , 2024

    2024 arXiv

  54. [62]

    Lmcache: A large language model powered intelligent cache system,

    LMCache, “Lmcache: A large language model powered intelligent cache system,” GitHub repository, 2023, accessed: March 2025. [Online]. Available: https://github.com/LMCache/LMCache

    2023

  55. [63]

    Do large language models need a content delivery network?

    Y . Cheng, K. Du, J. Yao, and J. Jiang, “Do large language models need a content delivery network?” arXiv preprint arXiv:2409.13761 , 2024

    2024 arXiv

  56. [64]

    vllm production stack: Reference system for k8s-native cluster-wide deployment,

    vLLM Project, “vllm production stack: Reference system for k8s-native cluster-wide deployment,” GitHub repository, 2025, accessed: March

    2025

  57. [65]

    Efficient memory management for large language model serving with pagedattention,

    W. Kwon, Z. Li, S. Zhuang, Y . Sheng, L. Zheng, C. H. Yu, J. E. Gonzalez, H. Zhang, and I. Stoica, “Efficient memory management for large language model serving with pagedattention,” in Proc. ACM SIGOPS 29th Symposium on Operating Systems Principles , 2023. APPENDIX A. Sample ...

    2023

  58. [67]

    How are you feeling today, can you rate your overall health from 0 to 100, with 0 being the worst and 100 being the best? Patient: 25%

    Conversation Excerpt: Assistant: Hi Craig, this is Agent PULSE from Cleveland Clinic. How are you feeling today, can you rate your overall health from 0 to 100, with 0 being the worst and 100 being the best? Patient: 25%. Assistant: So that’s a 25 out of 100. I’m so sorry to h...

  59. [68]

    This example demonstrates how Agent PULSE can main- tain natural conversations while automatically extracting and structuring critical health information for healthcare providers

    Automated Analysis Output: The SOLOMON frame- work automatically analyzed this conversation and generated the following structured assessment scores: TABLE I MODIFIED HEALTH AND BEHAVIORAL INDEX (MHBI) A SSESSMENT Measure Score Notes Liquid Stools 3 Patient reported having dia...

  60. [69]

    The Basics: V oice AI healthcare systems must perform several critical functions:

    Non-Technical Summary: Understanding the Technical Foundations of Voice AI in Healthcare: For healthcare profes- sionals, it’s important to understand what makes voice-based healthcare AI systems work effectively and how potential improvements in these technical areas could di...

  61. [70]

    Listen to patients and understand their speech accurately

  62. [71]

    Remember previous parts of the conversation to maintain context

  63. [72]

    Generate helpful, accurate responses based on medical knowledge

  64. [73]

    Deliver these responses quickly and naturally

  65. [74]

    Research suggests this could potentially increase patient satisfaction and improve information disclosure rates

    Connect with existing healthcare systems to update records Why These Proposed Technical Improvements Would Matter for Healthcare: Improving how systems manage conversations (session management) could allow patients to experience more natural conversations with fewer awkward pa...

  66. [75]

    Technical Implementation Details

    Potential Technical Improvements and Patient/Provider Benefits: The following table connects proposed technical improvements with potential healthcare outcomes: C. Technical Implementation Details

  67. [76]

    frontend

    KV Cache Optimization for Efficient Conversation Man- agement: Current frameworks for LLM-powered conversa- tional agents often handle each interaction round as an in- dependent API call to the LLM, leading to inefficient use of computational resources. Unlike discrete web pag...

  68. [2025]

    Available: https://github.com/vllm-project/production- stack

    [Online]. Available: https://github.com/vllm-project/production- stack

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.