Can AI Provide Better Consultation Than Humans When There Is a Fetal Malformation?
AI won’t replace empathy, but it can strengthen it, and improve precision
A Case That Stays With Me
I still remember the day we diagnosed a cardiac malformation in a fetus at 20 weeks of pregnancy. The couple was expecting their first child. They had walked into the ultrasound room excited and smiling. Within minutes, their world changed. As I explained the diagnosis, the potential treatment options, and the uncertain prognosis, I watched their faces shift from joy to fear. They asked questions: Would their baby survive? Could surgery help? Should they prepare for comfort care?
I did my best to answer, but I knew the limits of that encounter. Twenty minutes is not enough to explain a lifetime of implications. I worried they might leave more confused than reassured. That day crystallized for me the challenge at the heart of prenatal counseling: how to provide families with accurate, comprehensive, and compassionate guidance when the stakes could not be higher.
All this changed with the introduction of AI models like Claude and ChatGPT.
Why Counseling Is So Difficult
The mid-trimester ultrasound (also known as “anatomy scan”), performed around 18–22 weeks of the pregnancy, is usually a joyful milestone for parents. Most scans are reassuring, confirming a healthy pregnancy and offering parents the first detailed glimpse of their baby’s development. Yet in about 2–3% of pregnancies, the scan reveals a fetal malformation. For some families, the finding may be relatively minor, such as an isolated small cleft lip or mild renal dilation, conditions often treatable after birth. But in other cases, the anomaly may be major and life-altering, such as major congenital heart disease, neural tube defects like spina bifida, or brain malformations such as holoprosencephaly. Cardiac anomalies alone account for nearly one-third of all prenatally diagnosed malformations, while central nervous system anomalies are detected in roughly 20–25% of cases. Skeletal, abdominal wall, and genitourinary defects make up much of the remainder.
When such a diagnosis is made, the tone of the visit changes instantly. What was meant to be a reassuring moment can become devastating, especially when the malformation carries significant risk of disability or death. Families must rapidly process complex medical information, weigh uncertain prognoses, and begin to consider decisions that may shape their child’s future and their own lives in profound ways.
Even with the best intentions, human-led consultations have limitations. Physicians differ in how they present information, leading to inconsistent messages. Time pressures in busy clinics compress long conversations into short visits. Personal bias may color how options are framed. And with thousands of rare conditions, even experts cannot recall every statistic, outcome, and care pathway.
The result is that families sometimes leave without fully understanding what has been found, what it means, and what options they truly have.
What Generative AI Can Offer
Generative Artificial Intelligence (AI) refers to models that can create new content—text, images, even diagrams—based on what they have learned from vast datasets. Unlike older forms of AI that only classify or predict, generative AI can produce customized explanations, summaries, and visual aids in real time.
Applied to prenatal counseling, this technology can:
Instantly synthesize up-to-date knowledge about even the rarest malformations.
Present structured, consistent explanations that cover causes, prognosis, treatment options, and complications without leaving gaps.
Adjust tone and language complexity to fit the family’s needs, whether highly technical or plain language.
Create visual diagrams and timelines that make complex anatomy or treatment paths easier to grasp.
Provide ongoing explanations and clarifications after the initial consultation, available anytime.
These abilities make AI a powerful partner in ensuring families receive accurate, consistent, and understandable information.
The Irreplaceable Role of Empathy
But facts alone are not enough. When parents hear their baby has a malformation, they are not only processing information—they are experiencing grief, fear, and uncertainty. What they need most in that moment is not just clarity but human presence.
Empathy means more than kindness. It is the ability to recognize the emotional weight of the diagnosis, to validate the family’s feelings, and to accompany them through uncertainty. Families remember not only what was said but how it was said. Tone, pacing, and the willingness to sit in silence together all matter.
AI can mimic compassionate words, and studies suggest that in some settings patients even perceive AI-generated language as more empathetic than human speech. This may be because AI never tires or becomes rushed. Still, a machine cannot hold a hand, share silence, or absorb the rawness of human grief. Empathy is ultimately relational and cannot be outsourced.
The future, therefore, is not about AI replacing empathy but about AI supporting it. Well-designed scripts and explanations can give clinicians the time and emotional space to focus on the human connection.
How AI Fits Along the Ultrasound Pathway
AI can play a role before, during, and after the 20-week scan.
Before the appointment, it can generate simple, easy-to-read handouts explaining what the ultrasound looks for and what to expect, helping reduce anxiety.
During the scan, AI can assist clinicians by suggesting possible diagnoses when unusual findings appear, providing structured checklists that ensure no detail is missed, and reminding providers of associated syndromes or additional features to evaluate.
After the scan, if a malformation is confirmed, AI can generate individualized summaries written in accessible language, annotated diagrams of the affected anatomy, and anticipatory guidance about next steps. Families can receive lists of questions to bring to specialists, translations in their preferred language, and support resources they might not otherwise know exist.
For example, in a case of mild ventriculomegaly, AI could explain the meaning of the finding, outline possible outcomes, emphasize the importance of follow-up imaging, and provide sample questions for parents to ask at their next appointment. In a devastating case such as anencephaly, AI could generate an explanation that is medically accurate yet gentle, acknowledging the emotional difficulty while outlining available options for care.
The Ethical Balance
AI’s promise also raises ethical questions. Respect for autonomy requires that families be given the best available information to make decisions. If AI can deliver more complete and unbiased counseling than humans alone, then ethically, families should have access to it.
At the same time, AI is not infallible. It can generate errors or “hallucinations.” This is why AI-generated content must always be reviewed by the physician. The clinician remains accountable. AI is a tool, not an authority.
Equity is another concern. High-quality counseling should not depend on geography, language, or hospital resources. AI has the potential to democratize access, but only if systems are designed to reach underserved populations as well.
A Hybrid Model for the Future
The best approach combines AI and human expertise. Imagine a consultation that unfolds like this:
AI generates a personalized packet with explanations, diagrams, and anticipated questions.
The physician reviews the material for accuracy and relevance.
During the meeting, the physician uses these resources as a foundation while offering empathy, context, and ethical guidance.
Families take home digital or printed materials they can revisit, reflect on, and share with loved ones.
This model gives families both the precision and comprehensiveness of AI and the compassion and wisdom of their doctor.
Looking Ahead
The real power of AI lies not in replacing physicians but in making physicians better. By handling the heavy lifting of information retrieval and structured explanation, AI frees doctors to do what only humans can do: connect, comfort, and guide.
Still, questions remain. How do we ensure accuracy with rare anomalies? How do we protect privacy when sensitive data is used? What governance structures will safeguard against misuse? And perhaps most importantly, will AI truly reduce decisional conflict and improve outcomes for families?
These are not reasons to avoid AI but to implement it carefully, with ethical guardrails and ongoing evaluation.
Conclusion
That couple I counseled years ago after their baby’s heart malformation deserved more than I could give in that one session. With today’s technology, I imagine a different outcome: a comprehensive, accessible packet tailored to their case, visual aids to support my explanations, and digital resources they could revisit together at home. And alongside these tools, I would still be there—to listen, to comfort, to guide.
So, can AI provide better consultation than humans when a fetal malformation is diagnosed? Not entirely. But it can make humans better. And in the service of parents, babies, and the physicians who care for them, that may be the most important advancement of all.