An elderly woman enters the clinic. She has been diagnosed with B-cell lymphoma and has yet to begin any form of therapy. The patient wants to know what her chances of recovery are, and to what extent they will improve if she undergoes chemotherapy. She also asks you if there is anything she can do in the meantime that will improve her chances, and help the chemotherapy to work. You consider the evidence in front of you and explain that the cancer was detected in its early stages, and with chemotherapy, about half of the patients can be cured. Your clinical experience, however, tells you that as she is in her 70s, her overall health is good. She quit smoking over a decade ago, and she has a strong support network at home. Her prospects could be better than average. Besides, you have known her for many years, and she has always been active, health conscious, and she already seems very positive and motivated, to beat her illness. On the other hand elderly the people tolerate the chemotherapy less well, and you don't know whether there is extra nodal spread. Perhaps there are ways that you can reach a more accurate and realistic prognosis for your patient. Prognostication is essential to daily clinical practice. It is vitally important that patients have an accurate idea of their future health, both for their well being and their own decision making. An accurate prognosis may also be crucial for clinicians to make the right judgments when it comes to deciding on the course of treatment or making recommendations for their patient. In fact, prognostication is a continuous process in medicine, with sequential updates following decisions. And all decision require some kind of consideration of a patient's health given a certain course of action. For our informal patient, the outcome of our prognostication will influence our treatment choice. The recommendations that we give to our patient. And it may influence the actions and quality of life of our patient. In order to reach an accurate prognosis, clinicians often require more than their own clinic experience. One approach could be to consider the pathophysiology of the disease in question and any knowledge of the disease mechanisms. However this kind of information is often quite limited and can be greatly influenced by extraneous factors. And what about healthy subjects? It is quite common to make recommendations for prophylactic interventions or lifestyle modifications in subjects who appear to be healthy, but at risk of a disease. In this situation, knowledge of disease mechanisms may have little value. Often instead we turn to general evidence that has been collected from a broad population, such as overall mortality rates, or disease incidences. But this kind of evidence is based on averages. And it likely our predictions will be wrong if our patient is different from the average patient. This can be seen as a relatively naive approach. And to aid us in reaching the correct prognosis for our patient, we need more detailed evidence that is specific to our patient. So instead we must turn to imperical clinical research. Looking into differences between patients. To help us make informed predictions about the future health of those patients. Ideally, clinicians aim to use information that is specific to that patient to make predictions about the course of health of that patient. Prognostic research aims to identify combinations of factors that can be used to translate information about individual characteristics of a patient, their disease and treatment into predictions of future health. This often accumulates into productions of tools such as prediction rules or scores to help clinicians convert information about that patient into meaningful and accurate predictions. While these kind of tools can help clinicians to reach more accurate predictions about future health of patient. We have to bear in mind that prognosis is a process steeped in uncertainty. As a result the tools that we developed based on our research will never be perfect, and can only provide the users with an estimated probability of something happening in the future. If we revert to use such a school for our informal patient, we might find that her estimated probability of survival after five years is 60%. For example, providing some guidance for our patient, but no definitive answers. It is important to note that the prognostic prediction should always be in terms of absolute values. It is not uncommon for research to provide patients and clinicians with relative information, such as relative risks or odds of an outcome over time. For example, a lymphoma patient might have a relative risk of mortality within 5 years of 0.8 compared to younger patients. But how to translate this into a prediction that makes sense to a patient management and clinical decision making. Thus prognostic research should always aim to produce tools for conditions that help them to estimate absolute risk for their patients. There are many kinds of clinical challenges that we can address with prognostic research. And many different kinds of information that in can provide conditions in their patients. We can use this information not only to have a clearer idea of the course of a patient's health, but to even assign some absolute risk of some health outcome over a period of time. Relative to another depending on treatment choice. Once we know a patient's risk, we can use that to categorize them as being at low, medium or high risk, which can be extremely helpful in making treatment decisions. Several guidelines now exist that give treatment recommendations for patients who are categorized into medium or high risk groups according to results from a prognostic scoring rule. Over the coming lectures we're going to discuss the key concepts in designing a prognostic study. The kind of information that need to be collected in the different ways of collecting that information that you can choose from. We will then go on to discuss how this information can be used to develop a prediction rule for use in clinical practice.