Lesson 1Foundations of sepsis pathophysiology and clinical criteria (SIRS, qSOFA, SOFA, Sepsis-3)This lesson reviews the biology of sepsis and how the body responds, then connects these processes to signs seen at the bedside like low blood pressure, fast heart rate, and organ problems. It compares SIRS, qSOFA, SOFA, and Sepsis-3 criteria and how they are used in sorting patients at emergency triage.
Body's response to infection and organ failureChanges in blood flow and small vessel problemsSIRS, qSOFA, SOFA: parts and limitsSepsis-3 meaning and criteria for septic shockWeaknesses of clinical scores in emergency careLesson 2Safety, failure modes, and mitigation: false positives/negatives, model drift, data quality issues, adversarial inputsThis lesson identifies safety risks like false alarms, missed cases, model changes over time, and poor data quality. It explores handling tricky or unexpected inputs, strong monitoring, safety barriers, human checks, and steps for safe model updates.
False alarms, missed cases, and ways they cause harmChecking data quality and spotting odd patternsModel changes, adjustments, and retrainingDealing with tricky or surprise inputsHuman checks, overrides, and management rulesLesson 3Evaluation metrics and validation strategies for sepsis prediction: AUROC, AUPRC, calibration, lead time, decision curve analysisThis lesson defines main ways to measure sepsis prediction performance, including AUROC, AUPRC, how well risks match reality, and early warning time. It explains checking inside and outside the study, time-based checks, and decision curve analysis to see real clinical value.
AUROC, AUPRC, and handling uneven classesRisk matching curves and grouping by danger levelEarly warning time and performance over periodsInside, outside, and time-based checksDecision curve analysis and overall gainLesson 4Feature engineering and temporal modeling: time-series preprocessing, sliding windows, trend extractionThis lesson explains cleaning and lining up emergency time-series data for models. It covers adjusting sample rates, dealing with uneven times, moving windows, features for trends and changes, and noting treatments and clinical background over time.
Lining up times, adjusting samples, and filling gapsMoving windows and prediction periodsTrend, change, and rate featuresNoting treatments and care steps upManaging uneven and thin time-seriesLesson 5Machine learning models for acute risk prediction: logistic regression, gradient boosted trees, RNNs, temporal convolutional networks, transformer-based time-seriesThis lesson compares ways to model sudden sepsis risk, from logistic regression to boosted trees and deep sequence models. It highlights strengths, weaknesses, clear explanations, and fit for emergency time limits.
Logistic regression and control choicesBoosted trees and feature weightsRecurrent networks for sequencesTime convolutional networks for seriesTransformers for clinical time dataLesson 6Data modalities for real-time sepsis detection: vitals, labs, nursing notes, medication, waveformsThis lesson describes main real-time data flows in emergencies, including vital signs, lab tests, medicines, nursing records, and body signal waves. It discusses how often sampled, reliability, and how each shows sepsis developing.
Vital signs and ongoing watch feedsLab sets, cultures, and wait timesMedicine orders, fluids, and pressure drugsNursing records, start checks, and chartsWaves from machines and bed toolsLesson 7Integration with ED workflows and EHR systems: event streams, FHIR, HL7, SMART on FHIR apps, CDS HooksThis lesson describes fitting AI sepsis models into emergency workflows and health records. It reviews event flows, HL7, FHIR parts, SMART on FHIR apps, and CDS Hooks, stressing ease of use, dependability, and little upset to clinical work.
Event-based setups and data flowsHL7 and FHIR parts for sepsis signsSMART on FHIR apps for bed decisionsCDS Hooks for aware suggestionsWorkflow fitting and ease testingLesson 8Clinical alert design and human factors: thresholds, alarm fatigue mitigation, escalating workflows, who receives alertsThis lesson covers rules for alert design for emergency doctors, including picking limits, stepped alerts, and sending to right roles. It addresses alert tiredness, timing, step-up paths, and showing reasons and background.
Picking limits and alert stepsAlert tiredness and cut-back plansWho gets alerts and on what linesStep-up workflows and shift supportExplaining alerts and giving backgroundLesson 9Regulatory and evidence requirements for diagnostic AI: FDA/CMS considerations, clinical validation study design, prospective pilots, reporting standards (TRIPOD, CONSORT-AI)This lesson outlines rules and proof needs for diagnostic AI in sepsis, including FDA paths, CMS thoughts, and clinical checks. It reviews forward pilots and report rules like TRIPOD and CONSORT-AI.
FDA paths for diagnostic help toolsCMS, payments, and quality plansPlanning strong clinical check studiesForward pilots and step rolloutsTRIPOD and CONSORT-AI report guidesLesson 10Deployment frequency and latency considerations: near-real-time streaming vs batch scoring, handling missing and delayed dataThis lesson discusses setup ways for sepsis models, comparing almost real-time flows with group scoring. It addresses time limits, handling missing or late data, filling back, and watching data line health in emergencies.
Almost real-time flows vs group scoringTime limits and service promisesFilling for missing and late inputsFilling back, replay, and late dataWatching lines and system strength
