back to impact
making sepsis detection more actionable for icu teams
designing a clinical experience that turned predictive signals into clearer, faster decisions at the bedside.
40%
reduction in time to treatment
industry
healthcare and life sciences
type
ai clinical decision support platform
mandate
critical care experience design
the platform
the platform used AI to identify early signs of sepsis and help intensive care teams intervene sooner.
but prediction alone was not enough. clinicians needed to understand what the system was seeing, judge its relevance quickly and act without leaving the workflows they already relied on.
clarient helped shape that intelligence into an experience built for the pace and pressure of critical care.
the clinical brief
clarient was brought in to rethink how predictive insight appeared inside the icu workflow.
the work covered research, workflow design, dashboards, alerts, prototyping, front-end development and emr integration. the experience was tested in icu environments, including johns hopkins and cleveland clinic, so design decisions could be evaluated against the realities of clinical work.
our contributions
User Research and InterviewsUX Strategy and Workflow DesignUI Design and Design SystemInteractive PrototypingReact Front-End DevelopmentPredictive Alert ExperienceClinician DashboardsEMR Integration
the decision gap
in an icu, more information does not automatically lead to a better decision.
clinicians are already working across vital signs, lab results, patient history and changing treatment plans. a predictive system had to surface meaningful risk without becoming another source of noise.
the experience also needed to earn trust. teams had to understand why an alert deserved attention, see the patient context around it and move towards treatment without breaking their existing workflow.
the design response
we began with the clinical workflow rather than the model.
research and testing helped identify where predictive information could genuinely help, how much detail clinicians needed in the moment, and where the interface had to stay quiet.
surface what needs attention
alerts were designed to bring meaningful changes forward without overwhelming clinicians with constant interruption.
the aim was to make urgency visible while keeping the surrounding patient context close at hand.
keep the evidence close
a prediction is more useful when clinicians can understand what sits behind it.
dashboards brought patient trends, treatment information and relevant clinical signals into the same view so teams could assess an alert without piecing the story together elsewhere.
keep action inside the workflow
the experience was designed to reduce the distance between seeing a risk and responding to it.
clinical actions, treatment tracking and feedback could happen within the platform while emr integration kept the experience connected to existing hospital systems.
surface what needs attention
alerts were designed to bring meaningful changes forward without overwhelming clinicians with constant interruption.
the aim was to make urgency visible while keeping the surrounding patient context close at hand.
keep the evidence close
a prediction is more useful when clinicians can understand what sits behind it.
dashboards brought patient trends, treatment information and relevant clinical signals into the same view so teams could assess an alert without piecing the story together elsewhere.
keep action inside the workflow
the experience was designed to reduce the distance between seeing a risk and responding to it.
clinical actions, treatment tracking and feedback could happen within the platform while emr integration kept the experience connected to existing hospital systems.
the impact
clarient helped turn predictive intelligence into an experience clinicians could use under pressure.
the interface brought patient information, trends and sepsis risk into a clearer working view. Instead of asking clinicians to interpret another stream of data, the product helped them see where attention was needed and understand the context around that signal.
predictive alerts were designed to support urgency without adding unnecessary interruption. customizable dashboards allowed different users to focus on the information most relevant to their role and the patient in front of them.
the experience also made feedback part of the product. clinicians could respond to ai recommendations within the workflow, giving the system more input while keeping that interaction close to the clinical decision itself.
responsive access across web and tablet supported teams working throughout the icu, while emr integration reduced the need to move between disconnected systems.
according to the project metrics provided, early sepsis detection outcomes improved by 18% and time to treatment fell by 40%. User feedback to the AI model increased by almost three times, giving the system a stronger stream of clinical input.
pilot sites also recorded 100% clinician adoption, with the experience tested in icu settings at johns hopkins and cleveland clinic.
the important shift was not simply that the ai could identify risk. It was that clinicians had a clearer way to understand that risk and decide what to do next.