Rethinking Rogers' Bell Curve: Innovation Diffusion Theory and the Challenges of AI Adoption in Healthcare

Everett Rogers' seminal work on diffusion of innovations (1983) defines diffusion as the process by which an innovation spreads through specific channels over time within a social system, emphasizing the uncertainty inherent in new ideas and technologies.

Rogers' classic bell‑curve model categorises adopters into innovators (2.5%), early adopters (13.5%), early majority (34.5%), late majority (34.5%) and laggards (16%), each passing through four cognitive stages: awareness, decision, initial use, and continued use, with five key attributes influencing adoption: relative advantage, compatibility, complexity, trialability and observability.

Since the 1960s the model has evolved: Foster introduced the technology S‑curve (1986), Christensen expanded it (1997) and highlighted the importance of timing and continuous innovation; Moore’s "Crossing the Chasm" (1991) identified a gap between early adopters and the early‑majority, describing the “death valley” that many startups must survive; and May et al. proposed Normalization Process Theory (2009) to explain how new practices become embedded in healthcare organisations.

The article then applies these theories to artificial intelligence in healthcare, noting that despite massive private investment, AI adoption remains limited due to five inter‑related factors: (1) mis‑applied consumer‑tech business models, (2) an extended “death valley” caused by long adoption cycles and talent scarcity, (3) poor fit with Rogers' five adoption attributes (low compatibility, high complexity, limited trialability), (4) divergent stakeholder groups that hinder communication, and (5) the lack of robust, drug‑like clinical trial standards for AI software.

Further, the discussion cites Christensen’s distinction between disruption (a process) and “break‑through” (a product), arguing that AI startups must continuously innovate to avoid becoming short‑lived “single‑product” firms, and that true disruptive impact in healthcare may require decades rather than the 15‑20‑year window typical of other technologies.

In conclusion, the article suggests that overcoming AI adoption challenges in healthcare will depend on aligning innovation diffusion theory with the specific regulatory, ethical, and resource constraints of the sector, and on redefining disruption as a sustained, iterative process rather than a single breakthrough.