Navigating Complex Healthcare Decisions in Cancer Treatment
Featuring: Michael Liebman, PhD
In short
Cancer diagnosis and treatment are complicated by the fact that disease is a constantly changing process, not a fixed state — meaning two patients with identical test results may actually be on very different paths. Dr. Michael Liebman, a computational biologist, walks through why test results can be incomplete or inconsistent, how to build a more productive relationship with your medical team, and how to think critically about AI tools and medical information. This is for anyone who wants to ask sharper questions and feel more confident in treatment decisions.
- •Keep a symptom and observation journal and share a copy with your doctor — it helps them see your disease as a trajectory over time, not just a snapshot at one visit.
- •Asking for a second opinion is reasonable and healthy; frame it to your doctor as wanting to feel fully confident, and ask if they can suggest a trusted colleague.
- •When your doctor gives you test results or a treatment plan, ask directly: 'How confident are you in this, and what are the uncertainties?' — you have every right to understand the limits of what the data shows.
- •AI tools and online medical literature can be a useful starting point for learning, but a recent study found fewer than half of high-impact medical research experiments were reproducible, so treat AI-generated information as a conversation starter, not a final answer.
Watch on Cancer Patient Lab YouTube
Ask anything about this — free, no signup
Instant answers grounded in real guidelines, not the internet.
Brad Power May 21, 2025 “There's ambiguity in diagnosing. If we take any kind of clinical variable or biomarker, what we can look at in three patients is that, at some point in time, patient #1 and patient #2 may look identical in one or multiple biomarkers, but the reality is, over time, because disease is a process, not a state, they may not be actually on the same trajectories.
Similarly, patient #1 and patient #3 are on exactly the same trajectories, but when they come in for diagnosis, they come in at different stages of the disease, so their test results are not the same. ” – Michael Liebman, PhD “Increasingly, we try to take advantage of AI, ML, and large language models to read the literature and give us some perspective of what's going on.
However, in a recent study of high impact papers, fewer than half of the experiments were reproducible. In other words, the data that's being used to generate these large language models when they aren't highly curated can be misleading and needs to be further tempered, but they can be a good starting point when care is used.
” – Michael Liebman, PhD “One sensitive area when it comes to trust is at the beginning of the whole medical journey, when you're trying to select an oncologist or a physician or some part of the medical team. A lot of people are uncomfortable and concerned that they will offend their current doctors by requesting to have a second opinion.
Meeting Summary
Cancer patients and caregivers face challenges in coordinating at least three complex systems: you (the patient), your disease, and the practice of medicine.
It is important to understand that disease is a process and not a state, and that complicates its diagnosis and potential management, especially since much of the critical data about you and about your disease may be missing, inaccurate, or not yet identified (measured), and the true understanding of disease continues to evolve.
Accurate and transparent communication with your medical team is critical to optimizing disease management and your outcomes. A basic understanding of the process of diagnosis, the challenges of clinical trials, and selection of treatment can lead to identifying the right questions for you to ask as well as how to evaluate and interpret the many channels of information.
Increasingly, another wrinkle is the possible use of AI In diagnosing and treating “your cancer”, and how you can determine what information you can trust. Are analyses based on “more data” better than those only using “good data”? How can biases, known or unknown, affect your confidence in your decision-making? Michael N.
D (theoretical chemistry and protein crystallography) is uniquely qualified to lead a discussion on the complexities of treatment decision-making. He is the Managing Director of IPQ Analytics, LLC, after serving as the Executive Director of the Windber Research Institute from 2003-2007.
He is an Adjunct Professor of Pharmacology and Physiology, Drexel College of Medicine, Resident Professor of Biology, University of Massachusetts-Lowell, and Adjunct Professor of Drug Discovery, Fudan University. Previously, he was Director, Computational Biology and Biomedical Informatics, University of Penn Cancer Center.
He served as Global Head of Computational Genomics, Roche Pharmaceuticals and Director, Bioinformatics and Pharmacogenomics, Wyeth. He was Associate Professor of Pharmacology and Physiology/Biophysics at Mount Sinai School of Medicine. He is an Invited Professor, Shanghai Center for Bioinformatics Technology, and of the Chinese Academy of Sciences.
He focuses on computational models of disease that stress risk detection, disease process, and clinical pathway modeling, and stratification from the clinical perspective. He utilizes systems modeling to represent risk/benefit analysis in pharmaceutical development and healthcare.
Current applications focus on women’s health: triple negative breast cancer, hypertension, and hypertensive disorders of pregnancy, infant-maternal morbidity and mortality, perimenopause-menopause transition addressing health disparities. He has launched a non- profit to focus on these women’s health issues. Why do you need to pay attention to how you make medical decisions?
•To improve the accuracy and personalization of your treatment, ultimately leading to better health outcomes
•To integrate and align multiple interconnected factors - you, your disease, and medical practice
•Because your physician has limited time to make decisions
•Because there are psychological biases in decision-making that can lead to errors, as highlighted by Nobel Prize winner Daniel Kahneman's work on slow and fast thinking processes What are key challenges in making complex medical decisions?
•Disease complexity : Diseases are processes, not static states, with varying trajectories and progression rates that are difficult to capture.
•Biomarker limitations : Current biomarkers often provide incomplete or inconsistent information, and their interpretation can vary between institutions. They represent measurable entities that we try to associate with our limited understanding of disease trajectories.
•Comorbidities: Patients often have multiple conditions that interact and complicate diagnosis and treatment.
•Heterogeneity: Patients receiving the same diagnosis can have very different underlying disease characteristics. What limitations in information from tests should you be aware of that can impact your medical decision-making?
•Biomarkers can be misleading because they may vary over time during disease progression, and different institutions may use different thresholds for positive/negative results.
•Test results can vary because different lab equipment and calibration can produce different results, different pathologists may interpret the same sample differently, and normal ranges are often based on population averages that may not reflect you.
•Test interpretation can be hard because tests don't capture the full complexity of your disease trajectory, comorbidities can significantly impact test interpretation, and discrete measurements might miss important trends or outliers. How can you make better medical decisions?
•Consider your disease as a dynamic trajectory rather than a static state
•Collaborate with researchers and clinicians to uncover deeper insights
•Focus on personalized approaches that recognize your variations in disease progression, lifestyle, and environmental factors How can you create a collaborative, two-way dialogue with your medical team so that they understand your unique situation and concerns and you feel fully informed and engaged in your care?
•Keep a detailed journal of symptoms, observations, and questions, and share a copy with your physician, but don’t expect them to read it during your office visit. Ask probing questions about your specific condition, such as "Are there other perspectives or approaches we should consider?"
•Request clarity on your physician's level of confidence and any uncertainties in the diagnosis or treatment plan
•Seek a second opinion respectfully, framing it as a desire to be fully informed and confident in the treatment approach
•Use nurse navigators as bridges to help translate complex medical information
•Focus on the four key elements of trust – consistency, compassion, communication, and competency – when selecting and working with your healthcare providers How can AI help in making complex medi medical decisions?
•Reading and synthesizing large volumes of medical literature
•Identifying patterns in complex disease processes
•Supporting more personalized approaches to diagnosis and treatment What are some limitations of current AI tools?
•Large language models can be misleading if not carefully curated.
•Fewer than half of high-impact medical research experiments are reproducible.
•AI should be used as a starting point for investigation, to enhance understanding, not as a definitive source for clinical decisions, or a replacement for human expertise and clinical judgment. What are the benefits of seeking second opinions?
•Confirm your diagnosis and treatment plan
•Explore alternative treatment options
•Gain additional insights into your condition
•Increase confidence in your medical decisions How can you navigate getting a second opinion without offending your current medical team?
•Approach your current doctor by saying you want to be proactive about your health
•Ask if they can recommend a colleague for a second opinion that they would trust
•Frame it as wanting to ensure you're exploring all potential options
•Emphasize that you value their expertise and are not challenging their judgment A good physician should support your desire to be fully informed and engaged in your healthcare. If a doctor reacts negatively to a request for a second opinion, that may be a red flag indicating you might want to seek a more patient-centered provider. How can you learn more about making medical decisions?
•See our previous conversation with Michael Liebman "Modeling Disease"
•Contact Michael Liebman at Michael.Liebman@IPQanalytics.com
•Share your treatment needs and preferences and a journal of observations and symptoms with your medical team and ask questions
•See other conversations on cancer care decision-making:
•• • •
•The information and opinions expressed on this website or platform, or during discussions and presentations (both verbal and written) are not intended as health care recommendations or medical advice by Cancer Patient Lab, its principals, presenters, participants, or representatives for any medical treatment, product, or course of action. You should always consult a doctor about your specific situation before pursuing any health care program, treatment, product or other course of action that might affect your health. For the video recording of this conversation, please see here. Meeting Notes KEYWORDS Women's health, computational biology, breast cancer, decision making, trust determinants, healthcare challenges, precision medicine, disease modeling, biomarkers, patient care, second opinions, clinical trials, lifestyle medicine, AI and ML, patient engagement. SPEAKERS Michael Liebman (71%), Chris Apfel (12%), Lea Ann Biafora (4%), David Plunkett (3%), Matthew DeAngelis (3%), Cindy Ness (2%), Roger Royse (2%), Mark Taylor (2%) CHAT CONTRIBUTORS Rick Davis, Chris Apfel, David Plunkett, Mark Taylor, Ari Akerstein, Lea Ann Biafora, Allen Morris, Roger Royse SUMMARY Dr. Michael Liebman discussed the complexities of healthcare decision-making, emphasizing the importance of understanding disease processes and patient-specific factors. He highlighted the challenges in precision medicine, noting the need for better integration of disease understanding and treatment. Dr. Liebman stressed the significance of trust, consistency, compassion, communication, and competency in healthcare providers. He also addressed the limitations of large language models in medical decision-making, advocating for evidence-based resources like UpToDate. The discussion included the role of nurse navigators, the importance of second opinions, and the potential of personalized medicine and lifestyle factors in improving patient care. OUTLINE Introduction and Background
•Michael Liebman, PhD, introduced himself and his group, IPQ Analytics, which includes a nonprofit focused on women's health called Woven.
•He has a background in computational biology and genomics, having worked on the original HER2/neu test and as the global head of computational biology and genomics for Roche.
•He is not a clinician and does not provide clinical advice, focusing instead on decision- making and the importance of trust in healthcare. Challenges in Healthcare Decision-Making
•The concept of the "three-body problem" (patient clinical process, disease understanding, and medical practice) in healthcare explains some of the challenges in the US healthcare system compared to other nationalized health systems.
•Decision-making in clinical practice is more condensed and less research-based compared to the scientific method.
•It is important to understand the root causes of clinical issues faced by physicians.
•Diagnosing and treating diseases is co Complexity of Healthcare”
•Diagnosing and treating diseases is complex, especially in the context of biomarkers and disease processes. Accuracy vs. Precision in Medicine
•Accurate medicine and precision medicine are different; a better understanding of a disease is needed before developing a drug.
•Simple approaches to disease treatment are limited and need more comprehensive research.
•There are positive trends in breast cancer survival rates, but the incidence of breast cancer is increasing.
•There is ambiguity in diagnosing and treating diseases due to the complexity of disease processes and the need for better staging and monitoring. Impact of Comorbidities and Inflammation
•Comorbidities impact disease progression and treatment response.
•There are challenges in diagnosing and treating diseases when patients have multiple conditions.
•Inflammation plays a big role in disease processes and inflammation caused by disease needs to be separated from that caused by stress or other factors.
•Pathology needs to develop a better understanding of disease heterogeneity. Role of AI and ML in Medicine
•The use of AI and ML can be used to read literature and provide perspectives on medical research.
•Large language models have limitations and need further validation of their findings.
•Trusted resources like UpToDate should be used for clinical decision-making. Patient Engagement and Second Opinions
•Patients should keep a journal of observations and share it with physicians to enhance communication and decision-making.
•Patients should seek second opinions, especially when feeling nervous about a diagnosis.
•Getting second opinions without offending current doctors is hard.
•Patients should be proactive in managing their care.
•Nurse navigators can bridge the gap between patients and physicians.
•Concierge medicine and other models can improve patient care and physician-patient communication. Lifestyle Medicine and Personalized Approaches
•Lifestyle factors are important in disease management.
•Personalized approaches are needed.
•Epigenetic measurements can help understand the impact of environmental exposures and lifestyle factors on disease.
•Nutrition is complex.
•Granularity is needed in medical models.
•Personalized medicine has the potential to improve patient outcomes.
•Better communication between researchers and clinicians is needed. Reliability of Test Results
•There is inter-rater variability in pathology.
•Consistent testing and calibration is needed.
•PSA (prostate specific antigen) testing has limitations, while other blood tests and a prostate MRI offer potential benefits.
•Evidence-based medicine is important.
•Superior care beyond the standard guidelines is needed. Conclusions
•Communication between physicians, researchers, and patients is important to improve healthcare.
•Better understanding of clinical problems and the importance of addressing the root causes of disease is important.
•Patients should take control of their healthcare and seek second opinions when necessary.
Full transcript
Michael Liebman The slides will be available, and I'll give you my email if you want to contact me on other issues. I have a group called IPQ Analytics. It is a commercial group, but we also have a nonprofit called Woven, focusing on women's health. FemHealth, not FemTech. It's an R&D, where “D” is not development, but discovery. I worked on the original HER2/neu test [ National Cancer Institute (NCI) def ].
many of you may be familiar with, that was before the drug Herceptin [ NCI def] was developed. Part of my job was to figure out what to do with the test. I was global head of computational biology [ CLRN def] and genomics [def] for Roche. I was the director of the Windber Research Institute, which is a DOD- sponsored comprehensive breast cancer program, jointly with Walter Reed. I focus on systems modeling, disease modeling, and women's health.
I am not a clinician, and as stated before, I do not provide clinical advice. Nothing in here is intended to be clinical advice. We talk about decision-making. I would refer you to look at the slide and the background from Dan Kahneman, the Nobel Prize winner, about the difference between slow and fast processes of systems thinking. We have to take into account as we consider this in decision-making, the external factors that come into play.
As cancer patients or caretakers or researchers, you understand there are many other factors besides pure science. But what I'll try to focus on in terms of decision-making, starts to address the science part itself. I'll touch on trust. What's critical for any patient is understanding how we've started to go beyond social determinants of health to understand cultural determinants and also trust determinants [health determinants def WHO ].
These are very critical.
Want to learn more about your specific case?
Upload your medical records and ask Navis questions tailored to your diagnosis.