Early Engagement Is Good. Continuous Engagement Is Better.

Bristol Myers Squibb recently published a white paper on early patient engagement in drug development. At its core is a simple but important idea: patients, caregivers, and advocacy organizations should be involved before key decisions are made, not after. The paper argues that lived experience should help shape development strategy, trial design, recruitment approaches, communications, and patient support while programs are still flexible enough to change. It is a meaningful evolution from a model where patient perspectives were often gathered only after protocols were written and major decisions were already locked in.

I agree with that premise completely. But while reading the paper, I found myself thinking about something else. What if we’re still thinking about engagement as a series of events instead of a continuous signal?

Most patient engagement today happens in moments. A survey. A focus group. An advisory board. A protocol review. These activities are valuable and necessary, but disease does not happen in moments. Caregiving does not happen in moments. A glioblastoma diagnosis unfolds over months and years, in thousands of small decisions and moments of uncertainty that rarely appear inside formal engagement exercises.

The moments that shape outcomes often happen after the appointment ends. They happen late at night when a caregiver is trying to understand a scan result. They happen when a family is weighing whether another treatment is worth the side effects. They happen when exhaustion starts influencing decisions more than optimism. They happen when someone quietly stops asking questions because they no longer have the energy to keep searching for answers.

The BMS paper highlights the downstream consequences of missing these experiences. Recruitment challenges. Participation burden. Retention issues. Protocol amendments. Disengagement. What stood out to me is that these outcomes rarely appear without warning. Long before a patient drops out of a trial, stops treatment, or disengages from care, there are signals. Confusion. Caregiver exhaustion. Navigation fatigue. Financial stress. Decision overload. Loss of trust. The outcomes are visible. The signals that created them often are not.

Healthcare has become very good at measuring what happened. We can measure biomarkers, imaging results, clinical outcomes, claims data, and electronic health records. We can see whether a patient stayed on therapy. We can see whether they completed a study. We can see whether they showed up for appointments. What we struggle to understand is why. The human factors that drive those outcomes often remain hidden.

For most of healthcare history, that limitation was unavoidable. Understanding patient experience required interviews, surveys, focus groups, and advisory boards. Those approaches remain important, but they are episodic by design. They rely on memory and retrospection. They capture snapshots of an experience rather than the experience itself.

Artificial intelligence creates a new possibility. For the first time, lived experience can become measurable. Not because algorithms replace human understanding, but because they allow us to identify patterns across thousands of conversations occurring in real-world settings. Signals such as caregiver burden, treatment anxiety, adherence risk, trial attrition risk, and quality-of-life decline can emerge long before they appear in traditional datasets.

This is where I think the conversation is heading. The next evolution is not simply earlier engagement. It is continuous engagement. Not replacing patient advocacy organizations. Not replacing patient experts. Not replacing advisory boards. Building on them. Creating systems that learn from patients and caregivers while experiences are unfolding instead of asking them to reconstruct those experiences months later.

At Ember, we’ve seen what happens when families have a place to turn between visits. Thousands of conversations and more than 44,000 messages reveal something traditional research methods often miss: the most important signals emerge outside clinic walls. Families return because the questions don’t stop after the appointment ends. Caregiver concerns appear early and persist throughout the journey. Treatment anxiety surfaces long before discontinuation decisions. Moments of uncertainty become visible before they turn into outcomes.

The most important insight may be that patients do not experience disease in milestones. Researchers organize disease into milestones. Clinical trials organize disease into milestones. Healthcare systems organize disease into milestones. Patients live it continuously. If we want research, development, and support programs to reflect reality, our ability to listen must become continuous as well.

The BMS paper is an important step forward because it recognizes that patient insight belongs upstream, where decisions can still be shaped. The next opportunity is recognizing that patient insight is not a single meeting, survey, or advisory board. It is a living signal. The organizations that learn how to listen to that signal will build better trials, better support programs, and ultimately better outcomes.

The patient voice has always been there. For the first time, we can hear it at scale.