Assignment_5_Multi-ClassificationDue_date_Mar_13th_2020

Assignment 5: Multi-Classification

Due date: Mar 13th, 2020 (Friday)

Total Points: 100

Please put your name, student ID, date and time here

Name:

Student ID:

Date:

Time:

In this assignment, you will investigate the handwritten digits dataset.

Sample images:

Please apply the folowing eight methods to classify the handwritten digits dataset.

Split the dataset into training sets and test sets

Fit the training data sets to the following eight algorithms

Print the classification report on the test data sets

Method 1: KNN

Method 2: Linear SVM

Method 3: Gaussian Kernel SVM

Method 4: Naive Bayes

Method 5: Decision Tree

Method 6: Random Forest

Method 7: Voting Classifier

Method 8: Bagging

Assignment 5 file:///C:/Users/Al-Ja/Downloads/Assignment 5.html

1 of 5 3/11/2020, 6:57 PM

In [4]: # Importing the dataset from sklearn.datasets import load_digits digits = load_digits() print(digits)

Assignment 5 file:///C:/Users/Al-Ja/Downloads/Assignment 5.html

2 of 5 3/11/2020, 6:57 PM

{'data': array([[ 0., 0., 5., ..., 0., 0., 0.], [ 0., 0., 0., ..., 10., 0., 0.], [ 0., 0., 0., ..., 16., 9., 0.], ..., [ 0., 0., 1., ..., 6., 0., 0.], [ 0., 0., 2., ..., 12., 0., 0.], [ 0., 0., 10., ..., 12., 1., 0.]]), 'target': array([0, 1, 2, ..., 8, 9, 8]), 'target_names': array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9]), 'images': array ([[[ 0., 0., 5., ..., 1., 0., 0.], [ 0., 0., 13., ..., 15., 5., 0.], [ 0., 3., 15., ..., 11., 8., 0.], ..., [ 0., 4., 11., ..., 12., 7., 0.], [ 0., 2., 14., ..., 12., 0., 0.], [ 0., 0., 6., ..., 0., 0., 0.]],

[[ 0., 0., 0., ..., 5., 0., 0.], [ 0., 0., 0., ..., 9., 0., 0.], [ 0., 0., 3., ..., 6., 0., 0.], ..., [ 0., 0., 1., ..., 6., 0., 0.], [ 0., 0., 1., ..., 6., 0., 0.], [ 0., 0., 0., ..., 10., 0., 0.]],

[[ 0., 0., 0., ..., 12., 0., 0.], [ 0., 0., 3., ..., 14., 0., 0.], [ 0., 0., 8., ..., 16., 0., 0.], ..., [ 0., 9., 16., ..., 0., 0., 0.], [ 0., 3., 13., ..., 11., 5., 0.], [ 0., 0., 0., ..., 16., 9., 0.]],

...,

[[ 0., 0., 1., ..., 1., 0., 0.], [ 0., 0., 13., ..., 2., 1., 0.], [ 0., 0., 16., ..., 16., 5., 0.], ..., [ 0., 0., 16., ..., 15., 0., 0.], [ 0., 0., 15., ..., 16., 0., 0.], [ 0., 0., 2., ..., 6., 0., 0.]],

[[ 0., 0., 2., ..., 0., 0., 0.], [ 0., 0., 14., ..., 15., 1., 0.], [ 0., 4., 16., ..., 16., 7., 0.], ..., [ 0., 0., 0., ..., 16., 2., 0.], [ 0., 0., 4., ..., 16., 2., 0.], [ 0., 0., 5., ..., 12., 0., 0.]],

[[ 0., 0., 10., ..., 1., 0., 0.], [ 0., 2., 16., ..., 1., 0., 0.], [ 0., 0., 15., ..., 15., 0., 0.], ..., [ 0., 4., 16., ..., 16., 6., 0.], [ 0., 8., 16., ..., 16., 8., 0.], [ 0., 1., 8., ..., 12., 1., 0.]]]), 'DESCR': ".. _digits_dataset:\n\ nOptical recognition of handwritten digits dataset\n --------------------------------------------------\n\n**Data Set Characteristic s:**\n\n :Number of Instances: 5620\n :Number of Attributes: 64\n :Attr ibute Information: 8x8 image of integer pixels in the range 0..16.\n :Missing Attribute Values: None\n :Creator: E. Alpaydin (alpaydin '@' boun.edu.tr)\n :Date: July; 1998\n\nThis is a copy of the test set of the UCI ML hand-written d igits datasets\nhttp://archive.ics.uci.edu/ml/datasets/Optical+Recognition+of+Ha

Assignment 5 file:///C:/Users/Al-Ja/Downloads/Assignment 5.html

3 of 5 3/11/2020, 6:57 PM

In [27]: import matplotlib.pyplot as plt digits.images[0].shape list = [10,100,100,45] fig = plt.figure() for i,j in enumerate(list):

plt.subplot(2,2,i+1) plt.imshow(digits.images[j],cmap='gray')

In [2]: X = digits.data y = digits.target

Step 1. Split the dataset into training data and testing data ( 10 points )

In [ ]:

Step 2. Algorithm Analysis ( 80 points )

Method 1. KNN

In [ ]:

Method 2. Linear SVM

In [ ]:

Method 3. Gaussian Kernal SVM

In [ ]:

Method 4. Naive Bayes

Assignment 5 file:///C:/Users/Al-Ja/Downloads/Assignment 5.html

4 of 5 3/11/2020, 6:57 PM

In [ ]:

Method 5. Decision Tree

In [ ]:

Method 6. Random Forest

In [ ]:

Method 7. Voting Classifier

In [ ]:

Method 8. Bagging

In [ ]:

Step 3: Accuracy Results Table ( 8 points )

KNN L_SVM RBF_SVM NB DT RF Voting Bagging

Accuracy

Weighted Precision

Weighted Recall

Step 4: Conclusion ( 2 Points )

In [ ]:

Assignment 5 file:///C:/Users/Al-Ja/Downloads/Assignment 5.html

5 of 5 3/11/2020, 6:57 PM

Chapter 13: Mobile Technology and mHealth

Robert Hoyt MD

John Sharp

Learning Objectives

After reviewing the presentation, viewers should be able to:

Describe the evolution from personal digital assistants to smartphones and the emergence of mHealth

List the various ways mobile technology is currently being used in healthcare

Compare and contrast mobile technology for clinicians and patients

Identify the limitations of mobile technology

Introduction

Mobile technologies, particularly smartphones, are extremely popular to all members of the healthcare team

Adding to the popularity:

Improved speed, memory, wireless connectivity and shrinking form factor (size and shape)

Affordable

Constantly improving features

Phone capability, email and access to Internet

A myriad of mobile apps for consumers and clinicians

Evolution of Mobile Technology

2G in 1990

3G in 2001

4 G in 2006

5 G ? 2020

Mobile health (mHealth) is the “medical and public health practice supported by mobile devices, such as mobile phones, patient monitoring devices, personal digital assistants and wireless devices”

Global Observatory for e-Health for the WHO (2011)

mHealth

Personal Digital Assistants (PDAs)

1990: Apple Newton $700 and bulky

1996: Palm Pilot

1999 Epocrates, free popular drug program

Later: PDAs with phone capability,

Internet access, WiFi

With huge consumer demand the transition to

smartphones was rapid by any standard

Smartphones

Defined as having an operating systems capable of hosting medical software or it Internet capable

Cloud computing allowed more medical programs of higher complexity to be accessed

Vast majority of adults and physicians carry a smartphone

Many access medical issues on the phone and a minority have at least one medical app and/or receive text messages from a healthcare system

The iPad was the first tablet to make an impact in healthcare

This very well liked platform has been used in exam rooms and the hospital to provide a light weight means of reaching the Internet and EHR access

The reality is that tablets have shortcomings as well, such that the actual impact of tablets in healthcare is largely unknown

Tablet PCs

m-Health Conceptual framework

Mobile Technology and Patients

Devices: smartphone or tablet PC with apps connected to Internet

Text Messaging or Short Message Service (SMS). Used for:

Appointment reminders

Education

Disease management

Behavior modification

Medication compliance

Laboratory results notification

Public Health – Immunization

SMS messaging more likely to be tried in developing countries

Mobile Technology and Patients (Software categories)

Personal health record

Telemedicine

Medication reminders

Fitness coach

Immunization guides

Disease management

Prevention guides

Diagnostics

Vital sign monitoring

Mental health

Connect with healthcare system

Mobile Technology to Track Health Habits and Physiological Signs

New movement (“wearable HIT” and “quantified self”)

New devices and sensors to monitor diet, exercise, sleep, heart rate, respiratory rate, oxygen level, skin temperature, hydration, etc.

Oriented towards patients

Communicate with smartphone via Bluetooth LE

Smart watches a new platform

It remains to be seen how fitness and home monitoring data can be uploaded, analyzed and archived in an EHR. Who is reimbursing for this?

Should the data be analyzed automatically by machine learning algorithms and posted on a patient dashboard?

Does PGHD change behavior? Probably not

Only a minority of health apps are successful and persist

Patient Generated Health Data (PGHD)

Mobile Technology and Clinicians

Smartphones synchronized with office or hospital. Popular, but raises significant security and storage concerns

Several EHR vendors offer a specific iPad software package for clinicians

Medical Software categories for clinicians: drug information, calculators, databases, immunization guides, medical resources, prevention guides, diagnostics, sensors, image viewers, journal access, Medline searches, monitoring, coding, medical translator, EHR access, telehealth, dictation and remote data collection

Apple

HealthKit: for iOS and Watch OS. Use APIs to integrate app with OS

CareKit: open source SDK for patient monitoring

ResearchKit: iOS app can be used for research

Android

Google Fit: SDK to build apps using APIs

ResearchStack: for research

Research Droid: research using Android smartphones

Software Development Kits (SDKs) for mHealth

Devices and apps which are used for treatment or CDS must have FDA approval

All others do not need FDA approval

Regulatory Requirements

Mobile Technology Challenges

Distraction: at work and everywhere else

Technical: inputting, screen size and interoperability issues

Security: need BYOD policies

Lack of quality control: Mobile App Rating Scale

Lack of evidence (low quality studies)

Will new sensors and devices be reimbursed by payers or will the patient have to pay?

Mobile technology has evolved at a blistering pace

The era of mHealth is here but too early to know what the impact on healthcare system will be

Healthcare-related apps are popular and available for all platforms, but are they used?

Enterprise integration of mobile technology is evolving; smartphones integrating with EHRs

Conclusions

Chapter 14: Evidence Based Medicine and Clinical Practice Guidelines

Robert Hoyt MD

William Hersh MD

After reviewing the presentation, viewers should be able to:

State the definition and origin of evidence based medicine

Define the benefits and limitations of evidence based medicine

Describe the evidence pyramid and levels of evidence

State the process of using evidence based medicine to answer a medical question

Recall important online and smartphone evidence based medicine resources

Describe the interrelationship between clinical practice guidelines, evidence based medicine, and electronic health records

Define the processes required to create and implement a clinical practice guideline

Learning Objectives

Information technology has the potential to improve decision making through online medical resources, electronic clinical practice guidelines, electronic health records (EHRs) with decision support, online literature searches, digital statistical analysis and online continuing medical education

Healthcare reform is dependent on quality healthcare and that is dependent on the best evidence available

Introduction

Revised 2000: is a systematic approach to clinical problem solving which allows the integration of the best available research evidence with clinical expertise and patient values

EBM Definition

Why EBM is important from the viewpoint of the National Academy of Medicine: “Patients should receive care based on the best available scientific knowledge. Care should not vary illogically from clinician to clinician or from place to place”

We are moving slowly from anecdotal evidence to randomized controlled trials

Importance of EBM

Current methods of keeping medically or educationally up-to-date do not work

Translation of research into practice is often very slow

Lack of time and the volume of published material results in “information overload”

The pharmaceutical industry bombards clinicians and patients every day; often with misleading or biased information

Much of what is considered the “standard of care” in every day practice has yet to be challenged and could be wrong

Importance of EBM

Continuing Medical Education (CME): we know this is often ineffective

Clinical Practice Guidelines (CPGs): will be covered later but CPGs may not produce change

Expert advice: much better than no information but not always evidence based

Reading: “like drinking water from a fire hose”. Need focus so the best and most pertinent content is read

Traditional Methods for Gaining Medical Knowledge

A clinical question is generated. Use PICO tools *

Seek the best evidence using an EBM resource

Critically appraise the evidence using tools we will cover later in the presentation

Apply the evidence to the patient, taking into account their values, preferences and circumstances

Example: For patients over age 65 with a suspected stroke or TIA, which is the preferred initial imaging technique (MRI or CT or both)?

The EBM Process to Answering a Clinical Question

* P for patient or problem. Define the patient or group; I for intervention or therapy of interest; C for comparison with other drug or placebo?; O for outcome of interest, mortality? Some add T for type of study = PICOT. We will focus on therapy questions

Validity:

Internal: Is the study believable? Look for apparent biases or errors in selecting patients, measuring outcomes, conducting the study, or analyzing the results

External: are the results generalizable to your population of interest? If you have a geriatric population results on teenagers might be irrelevant

Results:

Results should be assessed in terms of the magnitude of treatment effect and precision which we will comment on later

Terminology Used in Answering Clinical Questions

Therapy Question: does the addition of Plavix to aspirin reduce the incidence of future strokes?

Prognosis Question: will lowering the average blood pressure to less than 120/80 reduce the likelihood of a stroke?

Diagnosis Question: which is more sensitive and specific for detecting a heart attack, CPK or troponin?

Harm Question: how many patients will have renal insufficiency due to an ace inhibitor compared to a beta blocker?

Cost question: which is more cost effective to reduce hospital readmissions for heart failure, inpatient case management or home health nursing care?

Most Common Types of Clinical Questions

Evidence Pyramid

Better evidence

But fewer published

articles

Case reports/case series: collections of reports on treatment of patients without control groups; much less scientific significance

Case control studies: study patients with a specific condition compared with people who do not the condition; less reliable than randomized controlled trials

Cohort studies: evaluate and follow patients who have a specific exposure or receive a particular treatment over time and compare them with another group that is similar but has not been affected by the exposure being studied; the two groups may differ in ways other than the variable under study

Randomized controlled trials (RCTs): subjects are randomly assigned to a treatment or a control group that received placebo or no treatment; only difference between the two groups is the intervention being studied.

“Double blinded” – both the investigators and the subjects do not know whether they received an active medication or a placebo

RCTs are the gold standard design to test therapeutic interventions

Evidence Pyramid Studies

Systematic reviews: answer a focused question by including multiple RCTs. Several reviewers scan the medical literature with specific criteria

Meta-analyses: quantitative summary of systematic reviews using statistical techniques to combine the results of several studies, thereby leading to larger numbers (more statistical significance) and a wider range of patients

Evidence Pyramid Studies

GRADE: Grading of Recommendations, Assessment, Development and Evaluation is a common framework, but many others exist

Level 1: High quality evidence derived from consistent RCTs

Level 2: Moderate quality evidence inconsistent or less methodologically strong RCTs; or exceptionally strong observational evidence

Level 3: Low quality evidence, usually from observational studies

Level 4: Very low quality evidence from flawed observational studies, indirect evidence or expert opinion

Levels of Evidence

A study can be statistically but not medically significant. There are tools that shed more practical light on the results, such as NNT

The risk for people who are in the experimental group is called the experimental event rate (EER); whereas the risk for people who are in the control group is called the control event risk (CER)

The relative risk is the EER/CER or a ratio

The relative risk reduction (RRR) is the EER-CER/CER

The absolute risk reduction (ARR) is the EER-CER

Number Needed to Treat (NNT)

The NNT is calculated by dividing 1 by the ARR (or 100/ARR if the ARR is expressed as a percentage and not a fraction

NNT calculates the number of patients who need to be treated with X to prevent one Y, over a specific time

An example: drug A is associated with stroke in 5% of patients, compared to 7% of patients on placebo over a one year trial

RRR = 5%-7%/7% = 29%; ARR = 7%-5% = 2% (ignore minus signs)

NNT = 100/2% or 50 patients need to be treated for one year to prevent one stroke

Number needed to harm (NNH) is calculated by harm percentage divided into 100

Web site http://www.thennt.com/ devoted to NNT and EBM

Number Needed to Treat (NNT)

Low yield of excellent quality articles in general

Many repeated studies show different results

Results are very frequently inconclusive

Many studies are poorly designed, have inadequate number of patients, and have some bias

Many medical journals are now requiring registration of all clinical trials in public registries

Limitations of the Medical Literature

Different rating systems for evidence

Different conclusions drawn by different people

Reviews are very time intensive so reviews don’t exist for every medical condition

Some view EBM as “cook book medicine”

RCTs are expensive, so often sponsored by drug companies, introducing bias

Limitations of EBM

Health informatics studies are under the microscope:

Often small, non-randomized, non-controlled studies

Frequently based on before and after implementation studies

Technology studies are hard to double blind

Early studies failed to look at unintended consequences of technology; they only looked for benefit

Excellent cost analyses are frequently missing

Much more on this in the textbook

Evidence Based Health Informatics (EBHI)

EBM Resources

JAMA Evidence

Clinical Evidence

Cochrane Library

Cochrane Summaries

Evidence Updates

ACP Journal Club

Practical Pointers for Primary Care

Essential Evidence Plus

TripDatabase

OVID

SUMSearch

Bandolier

Centre for EBM

Best Bets

Evidence Based Health Care

Google (insert “evidence based”)

NNT web site

MDCalc has EBM calculator tools

EBM for mobile technology: MedCalc3000 calculators are web based and exist for mobile platforms

More EBM Resources

The Institute of Medicine in 1990 defined clinical practice guidelines (CPGs) as: “systematically developed statements to assist practitioner and patient decisions about health care for specific clinical circumstances”

CPGs are based on EBM and are generally embraced by most specialties around the world

While they are embraced they are not always utilized by physicians for a variety of reasons

Clinical Practice Guidelines

Panel of subject matter experts convene to evaluate an important medical problem (high cost, high risk, etc.), usually with a disease management team, possibly IT, epidemiology, etc.

Team reviews local issue and evidence and examines national evidence; balancing costs, benefits and risks

Team evaluates the applicability of the guideline to the local scenario

Leadership at the top and the medical and nursing staff are critical for acceptance in a healthcare system

Implementation details are important as well; will it be paper or electronic based? Part of EHR?

Clinical Practice Guidelines Steps

Practice setting: are the clinicians too busy or indifferent to new evidence?

Contrary opinion: do the experts agree?

Sparse data: is the evidence conclusive?

Expect low initial acceptance

CPGS can be too long, without a summary

Lack of local champions

CPGs may lack patient input with both the writing of the CPGs and the implementation

Incentives to adopt may be missing for clinicians

Barriers to CPGs

CPG Example Hypertension CPG by Kaiser

CPGs available for mobile technology

Web based CPGs (risk calculators)

ATP III

FRAX

Gail breast cancer risk assessment

Stroke risk calculator

EHR-based CPGs:

Library of CPGs embedded in the EHR

Order sets for new admissions

Electronic CPGs

National Guideline Clearinghouse

National Institute for Health and Clinical Excellence

Agency for Health Care Research and Quality

Health Team Works

Institute of Clinical Systems Improvement

CPG Resources

EBM is the pursuit of the best available evidence

EBM is based on a hierarchy or pyramid of evidence

There are multiple limitations of the medical literature and EBM

The average clinician would benefit from understanding EBM and having tools available

CPGs help standardize care based on EBM

CPGs are becoming a routine part of EHRs

Conclusions

blog6

Get help from top-rated tutors in any subject.