AI-enabled clinical decision support: Healthcare AI is expanding clinical decision support, but the equity implications diverge sharply

Overview

The report examines how AI is changing clinical decision support through a proprietary survey of roughly 80 U.S. healthcare professionals and analysis of adoption, workflow convergence, safety, monetization and valuation. It concludes that CDS demand should rise, but multiple tools can coexist: this supports a more constructive view of Wolters Kluwer’s UpToDate while leaving BofA cautious on Doximity’s ability to convert engagement into durable growth.

Core views

Healthcare provides a strong adoption backdrop because clinicians face both information overload and staffing pressure. Medical knowledge is thought to double every 60 days, versus 50 years in 1950, while a historical estimate suggested that reviewing one month of primary-care literature would require 26 days of continuous reading. A Doximity survey found 82% of more than 600 physicians felt overworked and 76% said staffing shortages and other pressures had compromised care during the prior 12 months. AI can improve efficiency and care: 80% of clinicians in an Elsevier survey expect it to become a critical assistant within 5–10 years, while a randomized study of 2,384 patients found AI-enabled CDS reduced medication errors by 49% and adverse drug events by 47%. Adoption has moved from experimentation into routine professional use. The AMA’s 2026 survey found 81% of physicians used AI professionally, up from 66% in late 2024 and 38% in 2023, while average use cases per physician more than doubled from 1.1 to 2.3. More than 85% of respondents to BofA’s own survey were already using AI, with more than half in pilot or limited use and nearly one third using it broadly. Use remains concentrated in practical tasks: literature and research summarization led at about 39%, documentation and note generation accounted for roughly 28–30%, and assistive diagnosis was lower at about 17%. A separate Doximity survey of 3,151 physicians found daily AI use rose from 47% to 63% in less than a year, literature search increased from 22% to 35%, and 69% of AI users reported better care and outcomes. Clinical decision support has consequently become a central AI use case rather than a marginal tool. More than 80% of BofA survey respondents considered CDS very important, roughly two thirds used it daily, about 80% expected their engagement to increase and 30% expected it to increase substantially. Planned or active health-system adoption reached 68% for ambient documentation and 43% for AI-based clinical documentation integrity or CDS in 2026, up from 42% and 27% in 2025. Vendor positions are nevertheless unsettled: 52% of large health systems, 57% of small systems and 67% of outpatient respondents were at least somewhat likely to switch AI-scribe vendors, reflecting ongoing evaluation of accuracy, workflow fit and economics. The product boundaries between CDS, ambient scribes and communication tools are converging. Abridge has added CDS through an UpToDate partnership; OpenEvidence has expanded into visits, dialing, coding and patient engagement; Doximity has moved from networking and tools such as Dialer and Scribing into CDS with Doximity Ask; and Microsoft is positioning Dragon Copilot as a workflow and distribution layer that accepts third-party CDS content. BofA estimates U.S. TAMs of US$5–10bn for ambient scribes and US$1.3–2.6bn for workflow and communication tools, based on annual clinician pricing of US$1,200–2,000 and US$240–360, respectively, across 4.2–5.2mn addressable clinicians. Outsell estimates global CDS TAM at US$3.3bn. These separate product opportunities could ultimately sit within a much larger US$100bn healthcare and pharma technology and information market as workflows consolidate. The survey depicts a multi-tool market rather than a winner-takes-all structure. Nearly 90% of respondents used UpToDate, approximately half used OpenEvidence and 33% used Doximity Ask; more than half named UpToDate as their most-used tool. Around two thirds used more than one solution, the average respondent had access to 2.1 tools, and Exhibit 46 reports 71% of all respondents using multiple CDS tools. Among physicians and physician assistants, that proportion reached 85%. Roughly half also used general-purpose AI such as ChatGPT for work. BofA’s channel checks suggest OpenEvidence and Doximity are suited to point-of-care referencing, whereas physicians continue to use UpToDate for deeper review and research. Monetization follows two broad paths. OpenEvidence uses a free, advertising-supported model that has accelerated physician adoption, although 42% of survey respondents were uncomfortable with advertising next to CDS answers. UpToDate primarily sells enterprise subscriptions, with about 80% of revenue coming from institutional customers, while Doximity is initially offering free or low-fee access and seeking enterprise licenses alongside advertising. Budget growth may be restrained: most surveyed decision makers expected CDS budgets to rise only 0–5%, and about half of respondents were unwilling to pay extra for AI functionality. BofA therefore expects the strongest monetization to accrue to integrated tools that deliver measurable workflow efficiency rather than stand-alone AI features. Safety and trust are especially important because CDS can directly affect diagnosis and treatment. A cited NOHARM study found that unreviewed AI recommendations could create potential for severe harm in as many as 24.6% of cases, primarily through omissions. Specialized clinical systems outperformed general-purpose models on safety-weighted measures: AMBOSS LiSA scored 86.2% ± 2.6%, Doximity Ask 84.5% ± 2.6% and OpenEvidence 80.0% ± 3.6%, versus 74.6% ± 3.9% for Claude Opus 5, 70.1% ± 4.4% for GPT-5.6 Sol and 62.60% for Gemini 3.1 Pro. OpenEvidence subsequently reported 87.2% for Darwin on the benchmark. Incorrect or incomplete responses were the leading survey concern, cited by 81%. BofA argues that authoritative content, EHR integration, patient-specific context, physician review, distribution and user experience may therefore matter more than model capability alone as frontier models improve. For Wolters Kluwer, the evidence is incrementally reassuring despite some erosion in direct usage. UpToDate has more than 3mn users, is present at more than 50,000 institutional sites and is estimated to serve 50–60% of U.S. hospitals. Expert AI had been rolled out to about 2,500 U.S. hospitals and health systems, or more than 90% of U.S. Enterprise Edition customers, by 6 August 2026, although only about 25% of surveyed UpToDate users were using it. Nearly 60% of surveyed decision makers were at least somewhat likely to renew, while fewer than 10% were unlikely to renew; only 10% of all users and 11% of UpToDate users viewed OpenEvidence as a full replacement. UpToDate also represents only about 1% of estimated U.S. hospital IT budgets, based on US$920mn of revenue against US$64.1bn of aggregate spending, reducing the incentive to remove it. The Wolters Kluwer downside is concentrated in direct-to-clinician sales and pricing. App monthly users have declined at double-digit year-on-year rates, but BofA believes app and web data primarily reflect direct subscriptions, which are about 20% of UpToDate sales and less than 2% of group revenue. Pricing historically contributed about half of UpToDate’s growth, so modest CDS budgets and weak willingness to pay for AI may slow growth. BofA trims 2027 Health division growth to 4% from 5% and reduces the division’s growth outlook by about 1 percentage point, reflecting falling Pro revenue and moderately slower Enterprise growth, partly offset by Learning, Research & Practice. Its model still has Clinical Solutions growing around 4% and the group growing 5–6% overall. BofA believes Wolters Kluwer’s valuation discounts a more severe outcome. The shares trade at about 12x P/E, roughly 50% below the five-year average, and at 9x CY26E EV/EBITDA versus 13.5x for RELX. Historically, each 1 percentage point improvement in underlying revenue growth corresponded to approximately 1.6x of EV/EBITDA rerating; this relationship suggests the market is pricing less than 5% group growth. A separate valuation check using an 8.5% WACC and unchanged 2026–2030 forecasts requires a -1% terminal growth rate to justify the prevailing share price. BofA reiterates Buy and sets a €95 price objective, or US$108.60, based on 11.5x CY26E EV/EBITDA and a 20% discount to peers. Doximity presents a more difficult trade-off. Its broader workflow suite is improving engagement: quarterly active workflow prescribers rose 30% year on year in 1Q27, about 415,000 users employed AI tools in the quarter ended 30 June 2026, AI-scribe users increased tenfold year on year and 165 health systems had adopted its AI offering. Yet no physician or physician-assistant respondent selected Doximity as the single CDS tool used most, implying that Doximity Ask remains complementary. Limited incremental revenue also leads BofA to suspect that Ask is free or nearly free to enterprise customers, with uncertain willingness to pay a market rate. Doximity’s core vulnerability is its dependence on pharma advertising, which represents about 90% of revenue. Its advertising revenue is driven by physician monthly users, attention share, ad load, CPM and the number of pharma modules. New competitors such as OpenEvidence and OpenAI create more destinations for physician attention and pharma budgets, while only about 6% of all surveyed advertisers and 5% of top-20 pharma respondents expected Doximity’s AI product to expand total spending on the platform; many instead expected reallocation from existing modules. Although BofA estimates a US$2.5–5bn U.S. application-layer TAM and says a 15–20% share could represent a US$500–750mn opportunity, enterprise licensing might cannibalize advertising if paying health systems demand an ad-free product. BofA analyzes three Doximity outcomes. In the first, Doximity remains a complementary tool and advertising grows below market for several years before normalizing, with some incremental enterprise revenue. In the second, workflow functionality consolidates around a few winners and Doximity is marginalized, causing lower engagement, CPM erosion, stagnant or declining advertising revenue and further valuation compression. In the upside case, Doximity becomes a top-one or top-two consolidated workflow platform, creates incremental point-of-care inventory and grows at or above the market. BofA considers sustained investment necessary to compete with better-funded peers: subscription revenue grew only 2–8% year on year over the last three quarters, EBITDA margins fell from the mid-to-high 50% range to the mid-to-high 40% range, and 54% of advertising survey respondents reported increased discounts or incentives, the highest level in more than two years. The report expects low-single-digit to mid-single-digit growth and low-to-mid-40% margins, below consensus FY28–FY29 margins of roughly 47–50%, and reiterates Underperform with a US$20 price objective based on about 9x CY26E EBITDA.

Analysis framework

BofA begins with the structural drivers of healthcare AI adoption, then uses third-party physician and health-system surveys to assess adoption, use cases and safety. It combines those sources with a proprietary July–August 2026 survey of roughly 80 U.S. users and decision makers, bottom-up TAM calculations, product and business-model comparisons, usage data, channel checks and company financial exposure. It then translates the industry findings into revenue, margin, scenario and valuation implications for Wolters Kluwer and Doximity.

Methodology notes

• Other

  Proprietary healthcare-professional survey

  AlphaROC conducted BofA’s July–August 2026 survey of roughly 80 U.S. healthcare professionals, split approximately evenly between CDS users and decision makers; the report uses it to assess usage, renewal intent, budgets, willingness to pay, competitive positioning and safety concerns.

• Other

  Bottom-up total addressable market sizing

  The report multiplies addressable clinician counts by annual per-clinician pricing to estimate TAM for ambient scribes and workflow tools, then compares those estimates with third-party CDS and healthcare-technology market sizes.

• Valuation methodsEV/EBITDA valuation

  Peer and historical EV/EBITDA valuation

  BofA compares Wolters Kluwer with RELX and its own historical growth-to-multiple relationship, while valuing Doximity at about 9x CY26E EBITDA and Wolters Kluwer at 11.5x CY26E EV/EBITDA.

• Valuation methodsDCF (Discounted Cash Flow)

  Implied terminal-growth valuation check

  Using an 8.5% WACC and unchanged 2026–2030 forecasts, BofA calculates that Wolters Kluwer’s prevailing price would require a -1% terminal growth rate.

• Other

  Doximity workflow-consolidation scenario analysis

  The report tests three outcomes—remaining complementary, becoming marginalized, or emerging as a leading consolidated platform—and traces each through engagement, advertising revenue, enterprise licensing, margins and valuation.

• Other

  Advertising revenue driver decomposition

  Doximity advertising revenue is analyzed as physician monthly active users multiplied by attention share, ad load, CPM and the number of pharma modules, clarifying how competition and workflow engagement affect growth.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Wolters Kluwer (WOLTF / WTKWY)

  Covered incumbent CDS provider; BofA reiterates Buy with a €95 (US$108.60) price objective.

  Strengths

  UpToDate has more than 3mn users, deep enterprise and EHR integration, more than 90% institutional renewal rates, trusted expert-authored content and a small share of hospital IT budgets.

  Weaknesses

  Direct app usage has declined, AI monetization appears difficult, CDS budget growth is modest and pricing historically supplied about half of UpToDate’s growth.

  Comparison

  UpToDate remains the most-used and most frequently selected primary CDS tool, while Wolters Kluwer trades at 9x CY26E EV/EBITDA versus 13.5x for RELX.

  Risks

  AI entrants, weaker Health growth, pricing pressure, cyber incidents, regulation, macro deterioration, Tax & Accounting competition and key-person risk.

• Doximity (DOCS)

  Covered healthcare workflow and advertising platform; BofA reiterates Underperform with a US$20 price objective.

  Strengths

  Large physician network, rising workflow engagement, 165 health-system AI clients, 30% year-on-year growth in active workflow prescribers and tenfold growth in AI-scribe users.

  Weaknesses

  Doximity Ask remains complementary, enterprise monetization is unclear, approximately 90% of revenue comes from pharma advertising and required AI investment threatens margins.

  Comparison

  Doximity Ask was used by 33% of respondents but by none of the surveyed physicians or PAs as their primary CDS tool; UpToDate and OpenEvidence were preferred primary tools.

  Risks

  Advertising budgets may fragment across new platforms, discounts may rise, enterprise licensing may cannibalize advertising, investments may exceed expectations and margins may remain under pressure.

• OpenEvidence

  Private competitor whose free, advertising-supported CDS model has gained rapid physician adoption and is expanding into enterprise partnerships.

  Strengths

  Strong monthly usage, free access, specialized clinical content and broadening workflow functionality.

  Weaknesses

  Enterprise monetization remains unclear, and 42% of surveyed respondents were uncomfortable with advertising beside CDS responses.

  Comparison

  It ranked behind UpToDate but ahead of Doximity Ask in survey usage and is generally viewed as a complement rather than a full replacement for incumbent CDS.

  Risks

  Health-system caution over accuracy, legal accountability, workflow integration and advertising suitability could slow adoption or monetization.

Key data

• Physicians using AI professionally81% in 2026Up from 66% in late 2024 and 38% in 2023.
• Average AI use cases per physician2.3More than double the prior 1.1 level.
• Daily CDS engagementApproximately two thirdsBofA proprietary survey.
• Expected CDS engagement growthApproximately 80%Share expecting usage to rise; 30% expect it to rise substantially.
• Use of UpToDateNearly 90%More than half selected it as their most-used CDS tool.
• Use of OpenEvidenceApproximately 50%Second to UpToDate in the proprietary survey.
• Use of Doximity Ask33%No physician or physician-assistant respondent named it as the primary CDS tool.
• Multiple CDS tools71% of all respondents; 85% of physicians and PAsThe average survey respondent had access to 2.1 solutions.
• AI response accuracy concern81%Respondents citing incorrect or incomplete answers as a leading concern.
• Estimated product TAMUS$5–10bn ambient scribe; US$1.3–2.6bn workflow/communication; US$3.3bn global CDSBofA estimates for the first two categories and Outsell estimate for CDS.
• UpToDate renewal intentNearly 60% likely; fewer than 10% unlikelySelected decision makers in BofA’s survey.
• Wolters Kluwer 2027E Health growth4%Reduced from 5% previously.
• Wolters Kluwer valuationApproximately 12x P/E and 9x CY26E EV/EBITDAP/E is about 50% below the five-year average; RELX trades at 13.5x EV/EBITDA.
• Doximity enterprise adoption165 health systemsAbout 415,000 workflow users used AI tools in the quarter ended 30 June 2026.
• Doximity revenue exposureApproximately 90% pharma advertisingCreates exposure to competition for physician attention and pharma budgets.
• Doximity operating outlookLow-single-digit to mid-single-digit growth; low-to-mid-40% EBITDA marginsBofA’s expected profile, versus consensus FY28–FY29 margins of roughly 47–50%.

Impact & implications

The report expects CDS to become more deeply embedded in clinical workflows, but not necessarily through a single dominant application. Multiple tools, enterprise distribution, EHR integration, authoritative content and demonstrable efficiency are likely to shape competitive outcomes more than model performance alone. For equities, that structure supports Wolters Kluwer because UpToDate’s enterprise presence, low share of hospital IT budgets and strong renewal intent may protect it from outright displacement, even if pricing and direct subscriptions slow. Doximity can gain engagement and enterprise exposure, but its heavy reliance on advertising means fragmented attention, pricing pressure and prolonged AI investment may prevent that engagement from producing durable revenue and margin growth.

Risks

• AI-generated CDS can produce incomplete or incorrect recommendations; the cited NOHARM study found potential for severe harm in up to 24.6% of evaluated cases without additional review.
• High switching intent shows that vendor positions remain fluid as health systems test accuracy, workflow fit and economics.
• CDS budgets are expected to grow only modestly, and roughly half of surveyed respondents are unwilling to pay extra for AI functionality.
• For Wolters Kluwer, direct-to-clinician usage and pricing may weaken even if enterprise contracts remain resilient.
• Wolters Kluwer’s stated downside risks include cyberattacks, data misuse, U.S. macro deterioration, regulatory change, Tax & Accounting competition, weaker Health growth, key-person risk and new AI competitors.
• For Doximity, rising competition for physician attention and pharma advertising could slow revenue and increase discounting.
• Doximity may need a longer and more expensive investment cycle, causing sustained margin compression.
• Doximity enterprise licensing may fail to command market pricing or may cannibalize advertising if paying institutions demand an ad-free product.
• Doximity execution risk is elevated by leadership changes and sales turnover during the reinvestment cycle.
• Doximity’s upside case would require stronger share gains, greater HCP-channel spending, better new-product penetration or more resilient margins than BofA currently assumes.

What to watch

• UpToDate enterprise renewal rates and whether the large undecided cohort ultimately renews.
• Actual usage of UpToDate Expert AI relative to its rollout across more than 90% of U.S. Enterprise Edition customers.
• CDS budget growth and willingness to pay for AI features, particularly whether increases remain in the 0–5% range.
• Whether declining UpToDate app and web usage remains confined to direct subscriptions or spreads to enterprise accounts.
• Health-system switching rates as AI buyers move from experimentation toward accountability and preferred-vendor standards.
• Accuracy, omissions and safety benchmarks for specialized CDS products versus improving frontier models.
• Doximity Ask’s conversion from free or low-fee adoption into incremental enterprise revenue.
• Whether Doximity Ask expands total pharma advertising budgets rather than reallocating existing Doximity spending.
• Doximity’s pricing incentives, subscription growth and progression toward the low-to-mid-40% EBITDA margin range.
• The pace at which CDS, scribes, EHR functions and communication tools consolidate into fewer workflow platforms.