Tag Archives: verbal autopsy

New VA paper from some colleagues

“Whenever they cry, I cry with them”: Reciprocal relationships and the role of ethics in a verbal autopsy study in Papua New Guinea
HN Gouda, A Kelly-Hanku, L Wilson, S Maraga… – Social Science & …, 2016 – Elsevier

http://www.sciencedirect.com/science/article/pii/S0277953616303318

Comments Off on New VA paper from some colleagues

Filed under Uncategorized

Mapping from SmartVA-Analyze output to ICD-10 causes

There has got to be a quick way to format this better:

Cause list for SmartVA against ICD-10 codes

ADULT CAUSES
Code to ICD-10 WHO ICD definition and comments
GBD Cause Group A: Communicable, maternal, neonatal and nutritional disorders
AIDS B24 Unspecified human immunodeficiency virus [HIV] disease
Diarrhea/Dysentery A09 Other gastroenteritis and colitis of infectious and unspecified origin
Malaria B54 Unspecified malaria
Maternal O95 Obstetric death of unspecified cause: Maternal death from unspecified cause occurring during pregnancy, labour and delivery, or the puerperium
Other Infectious Diseases B99 Other and unspecified infectious diseases
Pneumonia J22 Unspecified acute lower respiratory infection
TB A16 Respiratory tuberculosis, not confirmed bacteriologically or histologically
GBD Cause Group B: Non-communicable diseases
Acute Myocardial Infarction I24 Other acute ischaemic heart diseases (as for WHO 2014)
Asthma J45 Asthma
Breast Cancer C50 Malignant neoplasm of breast
COPD J44 Other chronic obstructive pulmonary disease
Cervical Cancers C53 Malignant neoplasm of cervix uteri (WHO VA has C55 for all female reproductive neoplasms)
Cirrhosis K74 Fibrosis and cirrhosis of liver
Colorectal Cancer C18 Malignant neoplasm of colon
Diabetes E14 Unspecified diabetes mellitus
Epilepsy G40 Epilepsy
Esophageal Cancer C15 Malignant neoplasm of oesophagus
Leukemia/Lymphomas C96 Other and unspecified malignant neoplasms of lymphoid, haematopoietic and related tissue
Lung Cancer C34 Malignant neoplasm of bronchus and lung
Other Cardiovascular Diseases I99 Other and unspecified disorders of circulatory system
Other Non-communicable Diseases R99 Other ill-defined and unspecified causes of mortality
Prostate Cancer C61 Malignant neoplasm of prostate
Renal Failure (due to renal disease) N19 Unspecified kidney failure
Stomach Cancer C16 Malignant neoplasm of stomach
Stroke I64 Stroke, not specified as haemorrhage or infarction
Other Cancers C76 Malignant neoplasm of other and ill-defined sites
GBD Cause Group C: Injuries
Bite of Venomous Animal X27 Contact with other specified venomous animals
Drowning W74 Unspecified drowning and submersion
Falls W19 Unspecified fall
Fires X09 Exposure to unspecified smoke, fire and flames
Homicide (assault) Y09 Assault by unspecified means
Other Injuries X58 Exposure to other specified factors
Poisonings (accidental) X49 Accidental poisoning by and exposure to other and unspecified chemicals and noxious substances
Road Traffic V89 Motor- or nonmotor-vehicle accident, type of vehicle unspecified
Suicide (intentional self-harm) X84 Intentional self-harm by unspecified means
CHILD CAUSES
GBD Cause Group A: Communicable, maternal, neonatal and nutritional disorders
AIDS B24 Unspecified human immunodeficiency virus [HIV] disease
Diarrhea/Dysentery A09 Other gastroenteritis and colitis of infectious and unspecified origin
Encephalitis G04 Encephalitis, myelitis and encephalomyelitis
Hemorrhagic fever A99 Unspecified viral haemorrhagic fever
Malaria B54 Unspecified malaria
Measles B05 Measles
Meningitis G03 Meningitis due to other and unspecified causes
Other Infectious Diseases B99 Other and unspecified infectious diseases
Pneumonia J22 Unspecified acute lower respiratory infection
Sepsis A41 Other sepsis
GBD Cause Group B: Non-communicable diseases
Other Cancers C76 Malignant neoplasm of other and ill-defined sites
Other Cardiovascular Diseases I99 Other and unspecified disorders of circulatory system
Other Defined Causes of Child Deaths R99 Other ill-defined and unspecified causes of mortality
Other Digestive Diseases K92 Other diseases of digestive system
GBD Cause Group C: Injuries
Bite of Venomous Animal X27 Contact with other specified venomous animals
Drowning W74 Unspecified drowning and submersion
Falls W19 Unspecified fall
Fires X09 Exposure to unspecified smoke, fire and flames
Poisonings X49 Accidental poisoning by and exposure to other and unspecified chemicals and noxious substances
Road Traffic V89 Motor- or nonmotor-vehicle accident, type of vehicle unspecified
Violent Death Y09 Assault by unspecified means
NEONATE CAUSES
Birth asphyxia P21 Birth asphyxia
Congenital malformation Q89 Other congenital malformations, not elsewhere classified
Meningitis/Sepsis P36 Bacterial sepsis of newborn
Pneumonia P23/J22 Congenital pneumonia/Unspecified acute lower respiratory infection
Preterm Delivery P07 Disorders related to short gestation and low birth weight, not elsewhere classified
Stillbirth P95 Fetal death of unspecified cause

Comments Off on Mapping from SmartVA-Analyze output to ICD-10 causes

Filed under global health

New Scientist did an article on VA

https://www.newscientist.com/article/2077535-diagnostic-app-can-reveal-cause-of-death-without-a-doctor/

Comments Off on New Scientist did an article on VA

Filed under global health

CCCSMF Accuracy

I have a new verbal autopsy paper out, which goes deep into the weeds of how to measure the accuracy of a method for identifying the underlying cause of death at the population level from survey data. http://www.pophealthmetrics.com/content/13/1/28

A fun online appendix includes probabilistic combinatorial calculations of the sort that I was actually trained in: http://www.pophealthmetrics.com/content/supplementary/s12963-015-0061-1-s2.zip

Comments Off on CCCSMF Accuracy

Filed under global health

Population Health Metrics Research Consortium Gold Standard Verbal Autopsy Data 2005-2011

One exciting announcement that I got to make at the Verbal Autopsy Congress last October is that the PHMRC gold standard verbal autopsy validation data is now available for all researchers. You can find it in the Global Health Data Exchange: Population Health Metrics Research Consortium Gold Standard Verbal Autopsy Data 2005-2011.

Insert an example of doing something with it here.

Comments Off on Population Health Metrics Research Consortium Gold Standard Verbal Autopsy Data 2005-2011

Filed under global health, machine learning

Random Forest Verbal Autopsy Debut

I just got back from a very fun conference, which was the culmination of some very hard work, all on the Verbal Autopsy (which I’ve mentioned often here in the past).

In the end, we managed to produce machine learning methods that rival the ability of physicians. Forget Jeopardy, this is a meaningful victory for computers. Now Verbal Autopsy can scale up without pulling human doctors away from their work.

Oh, and the conference was in Bali, Indonesia. Yay global health!

I do have a Machine Learning question that has come out of this work, maybe one of you can help me. The thing that makes VA most different from the machine learning applications I have seen in the past is the large set of values the labels can take. For neonatal deaths, for which the set is smallest, we were hoping to make predictions out of 11 different causes, and we ended up thinking that maybe 5 causes is the most we could do. For adult deaths, we had 55 causes on our initial list. There are two standard approaches that I know for converting binary classifiers to multiclass classifiers, and I tried both. Random Forest can produce multiclass predictions directly, and I tried this, too. But the biggest single improvement to all of the methods I tried came from a post-processing step that I have not seen in the literature, and I hope someone can tell me what it is called, or at least what it reminds them of.

For any method that produces a score for each cause, what we ended up doing is generating a big table with scores for a collection of deaths (one row for each death) for all the causes on our cause list (one column for each cause). Then we calculated the rank of the scores down each column, i.e. was it the largest score seen for this cause in the dataset, second largest, etc., and then to predict the cause of a particular death, we looked across the row corresponding to that death and found the column with the best rank. This can be interpreted as a non-parametric transformation from scores into probabilities, but saying it that way doesn’t make it any clearer why it is a good idea. It is a good idea, though! I have verified that empirically.

So what have we been doing here?

7 Comments

Filed under TCS

Global Congress on Verbal Autopsy in 2011 open for abstract submission

Have you heard me say that Verbal Autopsy is a exemplary machine learning challenge? I think I say it about once a week.

Now there’s going to be a great forum for saying it. Read more here.

Comments Off on Global Congress on Verbal Autopsy in 2011 open for abstract submission

Filed under global health