Try   HackMD

醫療器材查驗登記教育訓練測驗解析

主辦單位: 財團法人醫藥工業技術發展中心
日期: 2020.07.28

Claim: 以下解答純屬個人上完課後認為最合適的解答，但不代表答案完全正確。

Q1. 陽性預測值(positive predictive value) 不會受到盛行率(prevalence)的影響。

(A) 是
(B) 否
Ans. (B) 否，陽性預測值(PPV)會受到盛行率的影響。
陽性預測值定義為: 被檢驗為陽性者，真的患病的機率，以公式呈現如下:

PPV=P∗Sensitivity(1−P)(1−Specificity)
(Where P 代表盛行率)

Q2. 陰性預測值(negative predictive value) 是表示檢測的結果為陰性時，病人沒病的機會。

(A) 是
(B) 否
Ans. (A) 是，陰性預測值(NPV)代表被檢驗為陰性者，其真的沒患病的機率，以公式呈現如下:

NPV=(1−P)∗SpecificityP∗(1−Sensitivity)+(1−P)(Specificity)

(Where P 代表盛行率)

Q3. 敏感度(sensitivity)高的體外診斷醫材，適合用來做為篩檢(screening)的診斷工具。

(A) 是
(B) 否
Ans. (B) 否，特異性高的體外診斷醫材，才是適合用來作為篩檢的診斷工具，因為特異性高代表可以排除無患病者的能力較高；反之，敏感度高的體外診斷醫材適合用來做為確診的診斷工具，因為敏感度高代表從患病者檢驗呈現陽性的能力較高。

Q4. 診斷醫材的陰性相似比(negative likelihood ratio)越大，表示診斷的能力越強。

(A) 是
(B) 否
Ans. (B) 否，陰性相似比代表檢驗為陰性時，假陰性比真陰性的比值，因此陰性相似比值越小，診斷能力越好，越不易出現假陰性的結果。陰性相似比(LR-)公式如下:

LR−=(1−Sensitivity)Specificity
(Where LR-代表negative likelihood ratio)

Q5. 一個具信效度的體外診斷醫材的臨床效能試驗設計須包含下面哪些要素?

(A) 有經過具有代表性的病人群測試過。
(B) 有獨立(independent)且使用盲法(blind comparison)的與一個既定“黃金”標準比較。
(C ) 用標準診斷工具(reference standard)做確診時，不知道指標診斷工具(index test)的結果。
(D) 以上皆是。
Ans. (D) 以上皆是，設計具信效度的體外診斷醫材臨床試驗應包括以下項目:

  1. 代表性(representative)
    • 是否經過具有代表性的病人群測過?
  2. 確認(ascertainment)
    • 標準診斷方式是否不受到試驗醫材的結果影響?
  3. 測量(measurement)
    • 有標準診斷方式，其與試驗醫材間的比較是否為獨立的且使用盲法的?
    • 各別檢測方法是否由不同人獨立判讀?
  4. 方法複製的可近性(Permit replication)
    • 研究方法描述詳細可以複製嗎?
    • 是否依照方法描述即可做出相同的結果?

Q6：Who should take responsibilities in the conduct of the clinical performance study?

(A) Investigator
(B) Sponsor
C) Lab technicians
(D) All parties involved in the study
Ans. (D) 所有人均應負責臨床試驗的執行，並且有各別的角色責任。

Q7：Which one of the following items can be waived by ethics committee?

(A) IVD medical device
(B) Informed consent
C) All of above
(D) All None of above
Ans. C)根據IVD檢驗試劑的特性及臨床試驗使用的檢體類型，該IVD檢驗試劑可能判定為免除審查或免除受試者同意書，而由IRB開立免審證明。

(A) Anticipated duration of the study.
(B) Number of subjects expected to participate in the study.
C) Information on when results of the study will be provided to the subject.
(D) Description of the test using a medical device under investigation.
Ans. C) 不一定要將實驗結果的告知時間提供給受試者，但是當研究成果可合理預期對可辨識之檢體提供者個人健康有重大影響時應告知。

Q9. Who is in charge of defining the roles and responsibilities of all parties including those of the sponsor, monitor, principal investigator and study team members?

(A) Sponsor
(B) Principal investigator
C) Regulatory authority
(D) Committee
Ans. (A) 由Sponsor定義各別參與人的角色，如Sponsor, monitorm principle investigator等等。ISO 20916:2019-5.1

Q10：Who should be the person communicating with the ethics committee?

(A) Investigator
(B) Sponsor
C) All of above
(D) None of above
Ans. C) Sponsor或PI都可以直接跟IRB聯繫與討論，但因為PI的臨床試驗都是送同一機構的IRB，因此通常都是PI與IRB聯繫。

Q11：Which type of analysis should be if all test results are classified into two categories?

(A) quantitative
(B) qualitative
C) semi-qualitative
Ans. (B) 檢驗結果為區分兩種類別的檢驗類型為定性實驗，如果是semi-qualitative類別，檢驗結果將表示出陽性的程度比例(有多陽)或陰性的程度比例(有多陰)。

Q12：Which statistic should be used in the statistical analysis of quantitative when the candidate test is compared with the standard method?

(A) sensitivity
(B) kappa statistic
C) overall present agreement
Ans. (A) 定性檢驗與Gold standard比較的分析方法是Sensitivity/Specificty，大致可以這樣分:

定性:

  • Sensitivity/Specificity: 與Gold standard比較時採用的分析方法。
  • PPV/NPV/Overall percent agreement: 與Non-gold standard比較時採用的分析方法。
  • kappa analysis: 當實驗判讀會牽涉人的猜測因素時所用的分析方法。例如Rapid Test是否出現色帶的判讀就有可能出現不同人有不同判讀結果。

定量:

  • 第一步先畫圖Scatter polt or Bland-Altman plot(比較偏差程度時才用)。
  • 做Correlation coefficient相關分析，用Pearson correlation(線性相關時使用)
  • 做回歸分析，用Deming regression, Passing-Bablok regression, Least square regression繪製回歸線，接著與y=x的線進行比較看篇bias有多少
  • 用Assumption，null hypothesis為兩條線的偏差值<10來驗證假說

Q13：Is the larger value of kappa statistic referred to the higher level of agreement between the candidate test and the comparative method?

(A) Yes
(B) No
Ans. (A) 是，當kappa值越大代表給不同人判讀結果一致程度越高。

Q14：Is the value of overall present agreement always between values of negative and positive present agreement?

(A) Yes
(B) No
Ans. (A) 是 (這題我不確定)

Q15：Which method should be performed at the first analysis of quantitative?

(A) 2x2 table
(B) scatter plot
C) Bland-Altman method
Ans. (B) 定量分析第一步先畫圖，所以scatter plot or Bland-Altman plot應該都可以 (題目有點怪)

Q16：『人體生物資料庫管理條例』是何時公布施行？

(A) 100年2月3日
(B) 101年2月3日
C) 99年2月3日
(D) 102年2月3日
Ans. C) 人體生物資料庫管理條例於民國99年02月03日公布，101年1月26日修正，最近一次修正為108年6月12日。

Q17：人體生物資料庫管理條例所定義的生物檢體下列何者為非？

(A) 自人體採集之細胞、組織、器官、體液
(B) 不包含經實驗操作所產生，足以辨識參與者生物特徵之衍生物質
C) 不包括動物檢體
(D) 不包括cell line
Ans. (B) 檢體定義: 指與人體分離之細胞、組織、器官、體液或衍生物質(含萃取出來的遺傳物質)，包刮剩餘檢體、採集自胎兒或屍體之檢體。

Q18：以代碼取代參與者姓名、國民身分證統一編號、病歷號等可供辨識之個人資訊，使達到難以辨識個人身分之作業方式，此行為稱為

(A) 加密
(B) 編碼
C) 去連結
(D) 遮蔽
Ans. (B) 編碼定義：以數字或英文字母等代碼，取代檢體提供者姓名、身份證字號、病歷號等可供辨識個人資訊之作業方式。；去連結定義: 於檢體編碼後，將代碼與檢體提供者可供辨識個人資訊之對照資料完全永久消除之作業方式。

Q19：下列何者不可成為生物資料庫的設置者？

(A) 政府機關
(B) 大學教授
C) 醫療或學術機構
(D) 法人
Ans. (B) 生物資料庫之設置者，以政府機關、醫療或學術機構、研究機構、法人為限。個人不得設置。

Q20：人體生物資料庫的主管機關是

(A) 衛福部
(B) 內政部
C) 行政院國家科學委員會
(D) 教育部
Ans. (A) 人體生物資料庫管理條例第一章第2條，本條例所稱主管機關，為衛生福利部。

Last changed by 
 
Atomic Notes
這裡是Eric Juo(小卓)無聊發起的小小空間，目的在於共享生活中學習到的知識及經驗，這裡是一個新手友善空間，不要求筆記寫的多完整，只希望學習到的東西可以在某個地方保存下來，並可以幫助其他人一起學習
0
1
477

Read more

RNA-Seq overivew slide

lectures and hands on tutorial

Apr 24, 2022
RNA-Seq Analysis

RNA-Seq stands for RNA sequencing. It is a technology for studying RNA species. In a cell, the majority of RNA sepcies is ribosomal RNA (rRNA) which accounts for 90% of total RNA. Conversly, messenger RNA (mRNA) only aocunts for 1-2% of total RNA. The central dogma of biology tells us that DNA transcribes into mRNA and then mRNA translates into protein to execute biological functions. mRNA is the only RNA species we know encodes protein seuqences. Studying the level of mRNAs enables us to understand the gene expression level and infer the protein expression level. Therefore, most of time we only interest in the level of mRNA, and wants to deplet rRNA from the total RNA. Preprocessing RNA samples before sequencing includes total RNA extraction, mRNA enrichment, fragmentation, first-strand cDNA synthesis, RNA strand digestion, second-strand cDNA synthesis, 3&apos;end repair, adenylation and adatper ligation. In total RNA extraction step, extracted RNA should not have signs of degradation. In mRNA enrichment, poly-T beads or poly-T columns are used to separate mRNA from other RNA sepcies, since only mRNA contains poly-A tails. mRNAs are fragmented by chemicals or ultrasonic into certain size range that fits for the sequencing capacity of sequencer. mRNA are primed with random hexamer and reverse-transcribed into first-strand cDNA, followed by RNA strand digestion and second-strand cDNA synthesis. For making strand-specific RNA-seq, TTPs are replaced with dUTPs, which serves as marks for the second-strand, in the second-strand cDNA synthesis step. Since the nature of polymerase synthesis, the double strand cDNA product lose several bases at the 3&apos; ends on each strand. The 3&apos;ends are repaired by dna end repair enzyme and then added a adenine. Y-shape adapters with 3&apos;-T overhang are ligated to the 3&apos;-A overhang of the double strand cDNA. To reduce the complexity in short read assembly, dUTP marked cDNA stands are digested using uracil-N-glycolas (UNG). The single strand cDNAs are then subjected to sequencing. In sequencing step, primer targeting adapter sequence aneal to one end of cDNA. Fluoresence-labelled ATPs, CTPs, GTPs or TTPs is added one at a time and photographed. Layers of photographes are analyzied by computer to interpret which nucleotide is added to the growing strand in each reaction. For paired end sequencing, the other primer targeting adatper on the other side is added and squenced again. The resultant are 2 separate files, one from forward strand, the other from reverse strand. Raw reads obtained from sequencer should undergo quality control before downtream analysis. Raw reads data are normally reported in fastq format. Fastq format stores 4 lines of information for each read. The first line is read name, the second line is sequence, the thrid line is a "+" sign, and the forth line is ASCII characters encypting qualtiy scores. For illumina sequencer, the quality socre is reported in Phred33 scheme. To obtain numberical quality score, ASCII characters need to be converted to numbers they represented in computer and minus 33. Quality score is used to repersent how confidence we can say this base is called correctly. A 40 quality score means 1 in 10000 chances this base is called incorrectly. Thus, higher quality score indicating higher chance this base is called correctely. Usually, we would want to trim off bases with quality score lower than 20. Low quality score base calling usually happens at the 3&apos; end of reads since the polymerase is getting weak attaching to the template and prone to add wrong nucleic acids. FastQC is a commonly used tool to give an overview of read quality. It reports per base quality, per sequence quality, per base nucleotide content, per sequence GC content, number of duplicated sequences, overrepresented sequences and adapter content metrices. Per base quality metric shows the distribution of quality score in box plot for each base location. The upper boundary of the yellow box is the third quantile of qualtiy scores, and the lower boundary of the box is the first quantile. The red line at the middle of box is the median value of quality scores. Per sequence quality metric averages base quality for each read and shows the distribution of per read quality. For good quality reads, the distribution of read quality should skews toward high quality side. Per base nucleotide content metric shows the frequencies of nucleic acid appear at each base location. In principle, RNA has been randomly fragmented, so the frequency of each nucleic acid should distribute evenly across all read length. However, RNA-Seq data is an exception, a research has found that priming with random hexamer in cDNA synthesis step seems to have certain selection for fragmented RNA sequences, which then results in not that random nucleic acid content at the 5&apos;end of reads. But these bases are still real bases in mRNAs, not artifacts, it&apos;s ok to put them into short read assembly. Per sequence GC content metric cacluate GC content for each read and report the distribution of per read GC content. GC content is like a fingerprint of a species, it has a constant value for a given species. This also applys to GC content of reads. The distribution of per read GC content should peaks at the value equal to that species&apos; GC content. If peak shifts away from the expected GC content, it can stem from foreign species contaminants in read dataset. Number of duplicated sequences metric shows how many reads are exact maches to each other. For RNA-seq data set, highly expressed transcripts usually repeatly sequenced resulting in many duplicated reads. Overrepresented sequences metric reports reads that appear repeatively and account for 0.1% of total reads. Adapters are the most likely sequences captured by this metric. Adapter content metric align reads with public adapters and report alignment hits.

Apr 15, 2022
三角函數

Fundamental Trigonometric Identities Power Reducing Formulas Fucdamental Trigonometric Identities Reciprocal identities (&#x5012;&#x6578;) $sec\theta=\frac{1}{cos\theta}$ $csc\theta=\frac{1}{sin\theta}$ $cot\theta=\frac{1}{tan\theta}$ $cos\theta=\frac{1}{sec\theta}$

Oct 11, 2020
FASTAQ format

FastaQ&#x683C;&#x5F0F;&#x662F;&#x7528;&#x4F86;&#x5132;&#x5B58;&#x6838;&#x9178;&#x5E8F;&#x5217;&#x53CA;&#x5176;&#x5C0D;&#x61C9;&#x9E7C;&#x57FA;&#x54C1;&#x8CEA;&#x7684;&#x8CC7;&#x6599;&#x5132;&#x5B58;&#x683C;&#x5F0F;&#x3002; &#x5E8F;&#x5217;&#x54C1;&#x8CEA; &#x5E8F;&#x5217;&#x54C1;&#x8CEA;&#x662F;&#x7528;&#x4F86;&#x8A55;&#x4F30;&#x6B21;&#x4E16;&#x4EE3;&#x5B9A;&#x5E8F;(NGS)&#x4E4B;&#x5B9A;&#x5E8F;&#x7D50;&#x679C;&#x53EF;&#x4FE1;&#x5EA6;&#x7684;&#x6578;&#x503C;&#xFF0C;&#x5176;&#x516C;&#x5F0F;&#x5982;&#x4E0B;: $$Q = -10logP$$ Where Q:&#x4EE3;&#x8868;&#x5E8F;&#x5217;(&#x9E7C;&#x57FA;)&#x54C1;&#x8CEA; P:&#x4EE3;&#x8868;&#x51FA;&#x73FE;&#x932F;&#x8AA4;&#x7684;&#x767E;&#x5206;&#x6BD4;

Aug 23, 2020
Read more from Atomic Notes
Published on HackMD