根据现有癌症疾病模型对未来癌症人数的预测：
the forecast of the future cancer number based on the current model of the cancer prevalence 基于当前癌症发病率模型的未来癌症发病率预测
英国公共卫生观察站协会创建了一系列流行率模型，旨在预测取消流行率。未来的生存趋势需要根据当前趋势进行假设和预测。此外，时间依赖性发病率和未来发病率的外显趋势的线性拟合。因此，预测模型表明，存活数量如下：（见图1）
The Association Of PUBLIC HEALTH UK observatories have created a number of the prevalence model, which is designed for the prediction of the cancel prevalence number. Future survival trends require assump tions and projections, based on current trends. In addition, epoch dependent incidence and a linear fit for extrapo lating trends in future incidence. As a consequence, the prediction model has indicated that the number of the survivals as follows: (see Figure 1)
如图1所示，随着时间的推移，肺癌的总趋势将下降；2015年的存活率超过30000人。此外，到2020年，肺癌存活率可能在25000到30000之间。此外，肺癌的存活率可能低至25000人。
As can be seen in the Figure 1, general trend of the lung cancer would decrease over the time; there were more than 30000 survivals in 2015. In addition, in 2020, the number of the lung cancer survivals could be in a category of 25000 to 30000. Furthermore, the number of the lung cancer survivals could be as lower as 25000.
从合同上看，20052015年结直肠癌呈上升趋势。这个数字在2015年超过12万。此外，结直肠癌的存活率可能高达14万。此外，结直肠癌的存活率可能超过16万。
On the contract, there is a upward trend in the colorectal cancer from 2005 to 2015. The figure is more than 120000 in 2015. Additionally, the number of the colorectal cancer survivals could be as much as 140000. Furthermore, the number of the colorectal cancer survivals could be more than 160000.
因此，癌症患病率为2.5万人，肺癌和结直肠癌16万人。As a consequence, the number of the cancer prevalence as the 25000 in the lung cancer and 160000 colorectal cancer.
介绍目前癌症的Prevalence Model
Explanation of the cancer prevalence model
20 years ago, people often hear a word is: "is really perfect Bayesian analysis in theory, but it is a pity that cannot calculate the result" in the process of practical application. Happily, the situation has been greatly improved. Today, more complicated model also can be through the Bayesian method for processing. This improvement has attracted many new people to join Bayesian research, but also reduce the feasibility of Bayesian method of "philosophy" of controversy.
Bayesian calculation mainly concentrated in the test expectation (posterior expectation) on the calculation of this calculation requires special from one dimension to thousands of dimensional integral. Another common Bayesian calculation type is the calculation of the posterior distribution mode (posterior mode).
Posterior distribution expectation of traditional numerical calculation method is a numerical integral, Laplace approximate calculation and Monte Carlo sampling. Numerical integration in medium dimension (maximum 10) is very effective, on the question of the latest development of visible. Laplace and other saddle points (saddle point approximation methods discussed refer to those remarks R.S tradesman (vignette). So far, Monte Carlo sampling is important use of traditional method to calculate the posterior distribution expect the most commonly used method. This method can calculate dimension problem, and have high calculation accuracy.
At present, has become a very popular Bayesian method calculation method. On the one hand, because it is the efficiency of processing is very complicated problem, on the other hand because of its programming method is relatively easy. Need to stress is that is not to say that the MCMC method has been completely replace the traditional way, in some special occasions (e.g., required accuracy), the traditional method but also has its advantages.
Bayes statistical forecast model is a kind of dynamic model as the research object of time series prediction method. When doing statistical inference, general model is: a priori information overall distribution information sample posterior distribution information can be seen that the Bayesian model not only takes advantage of the early data information, also joined the decision makers the information such as the experience and judgment, and combined objective and subjective factors, with more flexibility to the abnormal situation happened.
The Bayesian formula is an important formula in probability theory, it is mainly used for calculating the probability of complex events, and it is essentially the integrated use of formula of addition and multiplication formula. Appeared in the 17th century, the Bayesian formula from discovered to the present, has been deeply into many aspects of science and society. It is under the condition of the observed event B has occurred, to find the probability of each reason caused β in. The Bayesian formula has extensive application in the actual life; it can help people to determine a result of the most likely causes of events.
癌症的Prevalence Model的准确性以及Limitations
Introduction of the current cancer prevalence model
Bayesian analysis could be summarized as follow: Givenβ as two mutually exclusive events A1, A2,...An, and one in probability. Ifβhas, in turn, know events, but I don't know it because of the A1, A2,...An, event that appears in the An to appear at the same time, in this way, in the presence of event B has appeared, the conditions, and events Ai (I = 1, 2,... n) of the conditional probability problems, solve the problem with the following formula: if B1,B2,……….Bn is Ω’a segmentation, i.e. B1, B2, B3……..Bn is incompatibility and
i=1,2,……,n. The socalled conditional probability, it refers to under the condition of A certain event B occurs, the probability of another event A, written as P (A/B).
The bayesian formula can be explained as follows: suppose there are n two mutually exclusive "reason" A1, A2,... ,An can cause the occurrence of the same kind of "phenomenon" B, if the phenomenon has occurred, the bayesian formula can be used to calculate the due to An Ai = (j = 1, 2,..., n) caused by the chances, if we can find a Ai, makes:
P(Aj/ B)= Max{P(Ai/B)} 1≤i≤n
Aj is caused "phenomenon" B "reason" as much as possible. Life often encounter such A situation, event A has occurred, we need to judge the cause of A "why" this would require the use of Bayesian formula to determine the probability of A "reason" happened. Bayesian decision under incomplete information, for some unknown state with subjective probability estimation, then using the Bayesian formula to modify probability, finally reuse expectations and fixed probability to make optimal decisions.
The accurate of this model could be better than the other model in the design through the time line of the cancer prevalence. Also, based on the feature of the model, the limitation could be summarized as follow:
Limitation
In theory, BAYESIAN model of minimum error rate compared with other classification methods. But in fact is not always the case, this is because the BAYESIAN model assumes that the properties are independent of each other, between the established hypothesis in practical applications are often not (can consider to use clustering algorithm to the correlation between the larger attribute clustering), it brought certain influence to the correct classification of BAYESIAN model. More in the number of attributes or attribute correlation between larger, BAYESIAN model classification efficiency than the decision tree model. In the attribute correlation is small, BAYESIAN model the performance of the most good.
结合Projections of cancer prevalence 20102040谈谈How this will impact drug market in UK.(drug discovery, number of ,pharma company pipelines...ect)
In the current situation of the report of the cancer prevalence in United Kingdom from 2010 to 2040, the report has indicated that cancer incidence would increase; however, the cancer survival s would also increase over the years. As a consequence, the drug for the cancer survivals would largely demand in the drug market.
Form the structure of the age for the cancer prevalence, about 77% of the cancer prevalence would be lived at least 65 years, which are 8 times in 2008, so that the drug quality for the cancer would be a challenge. Under the burden of the cancer prevalence, with the treatment of the cancer, “the first year following diagnosis and the last year of life contain the highest levels of acute cancerrelated health service utilisation, but there is also a significant amount of usage in the period 1–5 years after diagnosis.”As the drug planning, the number of the supply would increase from the 15years from the treatment.
“The number of survivors in each of the time since diagnosis bands o1, 1–5 and X5 years will increase, but the number who are longterm survivors will increase at the fastest rate – by 2040, 69% of all survivors will be at least 5 years beyond diagnosis under this scenario, compared with 62% in 2009.”Therefore, drug is the most important to extend the life for the cancer survivals.#p#分页标题#e#
That means
