- A personalized patient support program tailors support, such as onboarding, reminders and nurse or care-team outreach, to groups of patients with shared needs.
- H2O Driverless AI can build patient segments and drop-off risk models from your own program data, with reason codes for each score.
- Treatment and dosing decisions stay with the patient's physician. The models tailor program support, not therapy.
- Patient data stays in the environment you choose, used only under your own consent and privacy terms.
One program, many different patients
Your patient support program serves people with different conditions, circumstances and reasons for stopping a therapy. A single outreach plan treats them all the same, and the patients who need help most can be the ones you hear from least.
Your teams hold enrollment records, contact history and program feedback, but the patterns across them are hard to see by hand. By the time a patient leaves the program, the moment to help has passed.
How personalized patient support programs work
Program data becomes patient groups, risk scores and suggested next steps. Your program team decides the outreach, and physicians decide the treatment.
Step 1: bring together the program data you are permitted to use
Start with enrollment, contact history, program milestones and feedback, under your consent and privacy terms.
Step 2: find patient groups with shared needs
Clustering and other unsupervised machine learning methods group patients by shared characteristics and program behavior. Your team names each group and decides what support it gets.
Step 3: flag patients who may leave the program
H2O Driverless AI trains a classification model on past program outcomes, and automatic feature engineering builds signals such as missed contacts and time since the last touchpoint. Each patient receives a risk score and the reasons behind it.
Step 4: explain and document
Reason codes show what drove each score, and documentation is written automatically while the model trains.
Where the physician still decides
Treatment and dosing decisions stay with the patient's physician. The models tailor program support, such as outreach and reminders, not therapy. Patient data stays in the environment you choose, used only under your own consent and privacy terms.
Naming groups your team can act on
A cluster is only useful once your team can act on it. Name each group in the language your outreach team already uses, such as new to therapy, lapsed refill or high support need, and set the touchpoints for each one. The same groups can feed a nurse line, a reminder program or a copay assistance outreach, so the segmentation work pays off across more than one channel.